Linguistic Approaches to Literature (LAL) Linguistic Approaches to Literature (LAL) provides an international forum for researchers who believe that the application of linguistic methods leads to a deeper and more far-reaching understanding of many aspects of literature. The emphasis will be on pragmatic approaches intersecting with areas such as discourse analysis, sociolinguistics, ethnolinguistics, rhetoric, philosophy, cognitive linguistics, psycholinguistics and stylistics.
Advisory Editorial Board Timothy R. Austin
Donald C. Freeman
Raymond W. Jr. Gibbs
Loyola University Chicago Northern Arizona University University of Toronto University of Southern California University of Fribourg University of California, Santa Cruz Tel Aviv University
University of Maine Sheffield Hallam University Lancaster University University of Birmingham Tel Aviv University University Centre Luxemburg University of Amsterdam
Volume 5 Directions in Empirical Literary Studies. In honor of Willie van Peer. Edited by Sonia Zyngier, Marisa Bortolussi, Anna Chesnokova and Jan Auracher
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Directions in empirical literary studies : in honor of Willie van Peer / edited by Sonia Zyngier ... [et al.]. p. cm. (Linguistic Approaches to Literature, issn 1569-3112 ; v. 5) Includes bibliographical references and index. 1. Literature--History and criticism. I. Peer, Willie van. II. Zyngier, Sonia. PN36.P44D57 2008 809--dc22 isbn 978 90 272 3337 0 (Hb; alk. paper)
© 2008 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Part I Theoretical and Philosophical Perspectives Section editor: Paul Sopčák 1. S tudying literature and being empirical: A multifaceted conjunction Uri Margolin 2. E mpirical research into the processing of free indirect discourse and the imperative of ecological validity Geoff Hall
3. Notes towards a new philology Donald C. Freeman
Part II Psychology, Foregrounding and Literature Section editor: Olivia Fialho extual and extra-textual manipulations in the empirical 5. T study of literary response Peter Dixon & Marisa Bortolussi 6. Foregrounding and feeling in response to narrative David S. Miall
8. Narrative empathy and inter-group relations János László & Ildikó Smogyvári
9. E ffects of reading on knowledge, social abilities, and selfhood: Theory and empirical studies Raymond Mar, Maja Djikic & Keith Oatley
10. I magining what could happen: Effects of taking the role of a character on social cognition Jemèljan Hakemulder
Part III Computers and the Humanities Section editor: Vander Viana n automated text analysis: Willie Van Peer’s 11. A academic contributions Arthur C. Graesser & Brent Morgan
12. C omputationally discriminating literary from non-literary texts Max Louwerse, Nick Benesh & Bin Zhang
14. Searching for style in Modern American Poetry David L. Hoover
16. C onsolidating empirical method in data-assisted stylistics: Towards a corpus-attested glossary of literary terms Bill Louw
Part IV REDES Project: The new generation Section editor: Milena Mendes mpirial evaluation: Towards an automated index 17. E of lexical variety Vander Viana, Natalia Silveira & Sonia Zyngier
19. Proper names in the translation of The Lord of the Rings Vladimir Yepishev
20. Threat and geographical distance: The case of North Korea Jan Prasil, Maria Dudusova & Jan Auracher
21. Th e apology of popular fiction: Everyday uses of literature in Poland Maciej Maryl
A matter of versifying: Tradition, innovation and the sonnet form in English Walter Nash
Introduction Yet philosophers should not be specialists. For myself, I am interested in science and in philosophy only because I want to learn something about the riddle of the world in which we live, and the riddle of man’s knowledge of that world. And I believe that only a revival of interest in these riddles can save the sciences and philosophy from narrow specialization and from an obscurantist faith in the expert’s special skill, and in his personal knowledge and authority… (Popper 1959/1980: 23)
This volume has been designed as a tribute to Willie van Peer on the occasion of his 60th birthday for all the work he has done in promoting the area of empirical studies. For a very good reason, as our readers will soon see, junior researchers in Ukraine renamed the area: “vanPeerical studies”. So, instead of a tie or a pen, we agreed the most appropriate gift would be a book that he could share with others. In putting this book together, we followed the same methodology he and Max Louwerse once used in a previous publication: “we contacted friends and colleagues from a range of disciplines, mostly from literary studies, linguistics and psychology, to probe their interest in such an enterprise. The spontaneous response we obtained was so overwhelming that the present volume could be brought together, containing studies from leading scholars in a variety of fields” (Louwerse & van Peer 2002: ix). The difference here, however, is that we cannot say it all started on a warm summer evening, as the four editors work in four different countries in three continents, with contrasting climates, different time zones and various conditions. All the contributors in this book have in some way or another worked with Willie van Peer, either as editors, collaborators, supervisees, or admirers of his work. We also invited budding researchers, who owe much to Willie van Peer’s teaching and inspiration in all the courses he has taught around the world and who find an opportunity here to publish their work with seasoned scholars. By doing this, we believe we follow one of Willie’s leading mottos: stimulate students to actively and fearlessly carry out research independently on an international level from an early stage in their studies onwards, with the possible prospect of becoming researchers within and beyond today’s academia. The aims of this volume are twofold: we present a multidisciplinary approach to the field of empirical studies of literature by bringing together perspectives from a wide range of areas, such as philosophy, sociology, psychology, linguistics and
literature, and scholars who work in different parts of the world. These papers stand as a clear demonstration of the great breadth of Willie van Peer’s interests and knowledge. In addition, the budding researchers who contribute in this volume illustrate how bonds can be created between both junior and senior scholars from all continents, a principle Willie van Peer strongly holds and which can be found in the tenets underlying the redes Project, which he co-founded in 2002 (see Zyngier, Chesnokova & Viana 2007; also see www.redes.de). Popperian at heart, Willie van Peer’s interest in empirical studies results from his inquisitive mind and challenging posture. In his seminal work, published more than twenty years ago (van Peer 1986), he concentrated his attention on exploring whether literary quality could be found in the linguistic structure of texts. This dissertation was already empirical in nature as readers’ reactions were surveyed in order to validate the arguments, inspired by a speech on the new linguistic approach to analyze literary quality delivered by Geoffrey Leech. Going beyond formal and structural approaches, Willie van Peer wanted to observe the effects of literature on readers rather than predicting it without empirical evidence. His students and colleagues still often hear him say that some concepts can only be grasped by assessing what the majority thinks. This is how he might have worded it: when it comes to interpretation or opinion of elusive concepts like “beauty”, it is much more reliable to ask 30 people on the street than one solitary expert. It has been his constant concern to be as distanced as possible from theories and ideologies in order to find out what real people think and what their attitudes are. His program can probably be best described as democratic, and it necessarily leads towards an interdisciplinary approach at the interface between literature and psychology. Perhaps his experience in Medical School may have influenced the way he sees research: an evidence-based process, the quality of which depends much on collaboration and team-work. This perspective explains the number of his joint publications. Next to the shift towards psychology, Willie van Peer has also insisted on the need for statistics and a quantitative approach to literary studies. Granting that each reader has individual characteristics, he holds that trends in text-reader interaction need to be observed in larger groups and that it is high time that the hypothetical models of readers should be replaced by real people, who account for what the majority actually say, think and do. In his production, Willie van Peer has been exceptionally prolific. So far, he has published 5 books, edited or co-edited 11 volumes of articles, authored over 165 articles and 30 reviews (see http://www.daf.uni-muenchen.de/personen/ professoren/van_peer/index.html). The range of his academic interests comprehends narrative studies (van Peer & Chatman 2001), the role of the literary canon in education (van Peer & Soetaert 1993), and the quality of literary texts
(van Peer 2008). His publications, which are all interdisciplinary in nature, have greatly impacted a variety of fields, including stylistics, text linguistics, thematics, narratology. He has delivered over 200 talks to international audiences in 27 countries, which have inspired dozens of young and already mature researchers to bravely take a similar scholarly path. And more of his works are on the way. In fact, Willie van Peer had already developed many of his concepts by the time the International Society for the Empirical Study of Literature (esl or igel – Internationale Gesellschaft für Empirische Literaturwissenschaft) was founded, providing him with “a broad highway to unite systematically all appropriate metatheoretical, empirical, philosophical, historical, linguistic, sociological and psychological researches, in order to present a modern multidisciplinary approach” (Halász 1995: 11). It was actually with Siegfried Schmidt at the helm that sociologists, psychologists and literary scholars founded the igel in 1987. As Schmidt confessed at the Round Table held during the tenth biennial igel conference in Munich in 2006, the label had been coined for political purposes and he himself never intended the Association to actually take an empirical turn. However, the statement that the Society aimed at promoting empirical research both by junior and senior scholars (see www.igelweb.org) was taken literally by scholars such as Willie van Peer, who played a major role in promoting cooperation and personal contact in all areas supported by igel. As a founding member, member of the executive committee and former President of igel, he has been able to contribute both on an individual and an institutional level to empirical studies. By empirical, and very much in the light of our epigraph, we mean “a kind of reasoning and a kind of research that is based on real evidence, that is, on evidence from the real world, which can be inspected by anyone … one that bridges the gap between the Humanities and the Natural Sciences, in the realization that both need each other for a better understanding of the world” (van Peer et al. 2007: 7). And there are many systematic ways of conducting this kind of research. Some of them are questioning, observing, experimenting. Many of the chapters in this book illustrate different ways of being empirical. The present volume focuses on how this approach to Literature and other cultural artifacts builds bridges between different areas. It consists of four sections, each focusing on a different aspect of els development and application. The articles in the first section, “Theoretical and Philosophical Perspectives”, provide different ways of understanding what empirical means and what it entails. Section II, “Psychology, Foregrounding and Literature”, represents the interface in which Willie has contributed most in his attempt to define foregrounding. Part III, “Computers and the Humanities”, illustrates how new technology has impacted the understanding of literary texts and how computational linguistics can work to illuminate readers and their experience with verbal art. Section IV collects works
by young researchers who have participated in the redes Project Willie van Peer has been supporting for so many years now and who unarguably can be called The New Generation. Each of these sections offer an Introduction which will help the reader see the links between the chapters. The book closes with an Afterword, a tribute to Willie van Peer by one of his many distinguished colleagues. Here, then, we present the state of the art in empirical studies of literature. Much of this could not have been obtained were it not for the efforts of Willie van Peer, whose academic generosity, relentless search for knowledge, and provocative posture have helped make history. Sonia Zyngier Marisa Bortolussi Anna Chesnokova Jan Auracher January 2008
References Halasz, L. 2002. Freudian text as a challenge for empirical study of literature. igel 2002. Proceedings. Papers from the 8th conference held at Pécs, Hungary, August 21–24. Louwerse, M. & van Peer, W. (eds) 2002. Thematics. Interdisciplinary Studies. Amsterdam: John Benjamins. Popper, K.R. 1959/1980. The Logic of Scientific Discovery. London: Unwin Hyman. Van Peer, W. 1986. Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm. Van Peer, W. & Soetaert, R. (eds). 1993. The Position of the Literary Canon in Education. The Hague: Bibliographia Neerlandica. Van Peer, W. & Chatman, S. (eds). 2001. New Perspectives on Narrative Perspective. Albany NY: SUNY Press. Van Peer, W., Hakemulder, J. & Zyngier, S. 2007. Muses and Measures: Empirical Research Methods for the Humanities. Newcastle: Cambridge Scholars Publishing. Van Peer, W. (ed.). 2008. In press. The Quality of Literature. Linguistic Studies in the Evaluation of Literary Texts. Amsterdam: John Benjamins. Zyngier, S., Chesnokova, A. & Viana, V. 2007. Acting and Connecting: Cultural Approaches to Language and Literature. Munich: LIT Verlag.
Theoretical and philosophical perspectives Introduction Despite the fact that Willie van Peer has been an unparalleled ambassador for the empirical study of literature (esl), the field still needs to clarify its position among its participating disciplines. Regarding sociology, psychology or philosophy, this situation should not seem surprising, since in these disciplines literature and the arts in general are one among many objects of investigation. The uncertain relationship within literary departments, however, weighs more heavily on esl and raises the question of how fully it can be integrated with traditional literary studies. The uncertain position of esl may partly be due to the multi-disciplinarity and multi-nationality of the enterprise and the ensuing quandary of conflicting epistemologies within esl. Gerard Steen (2003) has described these tensions and their history as involving three competing “paradigms:” a) the focus on the philosophical and (meta-) theoretical groundwork of esl by the social constructivist group around Siegfried J. Schmidt at Siegen University; b) the focus on methodology by Norbert Groeben and colleagues; and c) the applied research of a group of mostly North American psychologists (e.g., Art Graesser in this volume). Despite the fact that the latter grouping has of late most strongly shaped esl internationally, the tension remains and continues to complicate the relations between esl and its ‘sister’ disciplines. The papers in this section address these issues in different ways. In the first paper of this section, Uri Margolin proposes a model of esl in which not only these three pillars of esl, but also historical, institutional, and purely conceptual approaches to the study of literature (e.g., linguistics, narratology, etc.), all function as necessary, mutually dependent, and complementary rather than competing discourses within a comprehensive empirical study of literature. On the basis of Mario Bunge’s (1998) philosophy of science and Siegfried Schmidt’s (1980) systems theoretical model of literary communication, he lays a solid epistemological foundation for esl. The strength of Margolin’s contribution lies not only in providing a framework within which each of the current approaches to esl find their place as interacting components (he posits a network where others have focused on divergence), but also in integrating esl within the larger structure of literary studies per se. This is a timely and welcome paper. Assuming a challenging position, Geoff Hall’s chapter underlines Margolin’s rejection of esl as limited to experimental approaches. Hall argues that all too often experimental approaches to studying certain aspects of literary reading are
reductive to the extent that they fail to make important distinctions and provide only trivial insights. To illustrate his point, he draws on two of the most influential recent empirical studies of free indirect discourse, which he considers exemplary of the complexity of literary reading. According to Hall, a greater emphasis on ethnographic and qualitative studies is called for to do justice to the complexity of literary reading and to ensure that the criteria for ecological validity are met. In the third chapter of this section, Don Freeman appeals for a return to “rational projects: inter-subjective, explicit, consistent, open to challenge” (Margolin, this volume) in the aftermath of what he considers the damage that poststructuralist “theory” has done not only to literary studies, but also to the Humanities in general. He proposes a “New Philology”, which, like Margolin’s proposal, encompasses both the purely conceptual study of literature as semiotic objects and empirical approaches. As illustration, Freeman applies the theory of cognitive metaphor to a passage from Shakespeare’s Macbeth. Don Kuiken’s contribution in some way touches on each of the preceding three chapters and also already sets the stage for the following section on psychology and literature. Peter Zima’s (1999) assessment of Adorno’s aesthetic theory comes very near to the approach Kuiken takes: “[I]t will systematically oscillate between Kant's refusal to conceptualise art and Hegel's claim that a conceptual definition is possible” (191). Whereas, Margolin, Hall, and Freeman adopt a conceptualizing approach to literature and literary reading, Kuiken, without accepting the poststructuralist rejection of subjectivity, shares its concern for the temporality and “fragility of felt meanings” (Kuiken, this volume). He draws on continental and analytic aesthetics, phenomenology, and in particular Gendlin’s (1962/1997) discussion of expression in order to develop a theory of reading as expressive disclosure. Kuiken’s proposal paves the way for forms of empirical investigation that provide access to reading moments during which “freshly felt meanings” are articulated, as well as to reading moments that reinforce poststructuralist claims about the unsayable (Budick and Iser 1989). He presents aspects of a number of his empirical studies to support his notion that in literary reading expression and disclosure are cotemporal. Each of these chapters, then, contributes in its own way to some of the theoretical and philosophical discussions that concern the dynamics within esl as well as its place in literary studies. By addressing aspects ranging from meta-theory to minute discriminations in “felt meanings”, and, most importantly, by demonstrating the “organic” structure that philosophy, theory, and applied research share, the following four chapters create a meaningful background against which the volume as a whole unfolds. Paul Sopčák Section Editor
References Budick, S. and Iser, W. (eds) 1989. Languages of the Unsayable: The Play of Negativity in Literature and Literary Theory. New York NY: Columbia University Press. Bunge, M. 1998. Philosophy of Science. New Brunswick NJ: Transaction Publishers. Gendlin, E.T. 1962/1997. Experiencing and the Creation of Meaning. A Philosophical and Psychological Approach to the Subjective. Evanston IL: Northwestern University Press. Schmidt, S.J. 1980. Grundriss der empirischen Literaturwissenschaft. Braunschweig: Vieweg. Steen, G. 2003. A historical view of empirical poetics: Trends and possibilities. Empirical Studies of the Arts 21: 51–67. Zima, P.V. 1999. The Philosophy of Modern Literary Theory. London: Athlone Press.
Studying literature and being empirical A multifaceted conjunction Uri Margolin I begin with a distinction between rational, empirical and experimentally based types of theoretical claims or discourses. I go on to discuss the study of literary dynamics (or literary historiography) as a field of empirical activity involving the relations and correlations between texts and text models on the one hand and the contexts in which they occur on the other. This is illustrated with respect to literary production, mediation, initial reception and post-processing over time (Siegfried Schmidt’s four components of literary communication). I conclude with a list of theoretical and methodological tasks for current experimental as well as empirical studies of literature. Keywords: literary theory, methodology, empirical study of literature, literary historiography, experimental studies, literary dynamics
1. L iterary object and theory: The dual conceptual/empirical nature of each The total literary system consists of two equally indispensable and mutually irreducible components: literature as a set of given semiotic objects and codes (text types, genres, styles) and literature as the correlated set of historically occurring individual and collective situations, activities and practices bearing on these objects and codes. The scholarly study and theorizing about literature as pure semiotic objects and codes and in a purely conceptual way (linguistics, semiotics, possible worlds theories, narratological models and theories, etc.) is a rational and inter-subjective kind of discursive activity, which produces both theoretical and descriptive poetics. Being rational and systematic, it cannot be faulted as such. But its validity and validation alike remain within the purely conceptual, as in philosophy, while the concepts and distinctions it employs may exist only within the scholarly or professional discourse system. Studying just the verbal objects
reduces and sometimes even eliminates the empirical factor, yielding a poetics in the abstract (e.g., New Criticism theories of poetic meaning or Northrop Frye’s grand typologies). On the other hand, studying just the correlated activities and situations reduces our discipline to institutional social history (Bourdieu 1984) or to individual psychology of cognition, discourse processing, etc. One obviously cannot disqualify either kind of study as a major part of the total study of literature, but one could still say, paraphrasing Saussure, yes, but I want to do something else: I want to study the life of literature au sein de la société et de l’histoire, this formulation providing an intuitive sense of the nature and drift of the specifically empirical discourse about literature. Studying literature and being empirical consists accordingly of examining in each case the nexus between these two components, the semiotic and the actional. My view of esl is that it is precisely the study of the interrelations between texts and text codes and the contexts in which they occur. esl thus constitutes the pragmatic component of literary studies, consisting as it does of the study of the relations between texts and codes and their actual producers and users. Differently put, for esl literature consists of semiotic objects not in isolation but as produced, mediated, perceived and post-processed by human agents who themselves operate in a defined socio-cultural and historical situation. (I am obviously following here S. J. Schmidt’s basic model of literary communication. See Schmidt 1980 and Hauptmeier & Schmidt 1985). In other words, esl consists of the study of semiotic objects as they occur in acts and processes of cultural communication, hence as utterances or Kommunikate (token, type). And any activity of literary communication is essentially historical, a point to which I will return. Now, both basic kinds of investigation, the semiotic and the institutional, can and should be rational projects: inter-subjective, explicit, consistent, open to challenge, verifiable in some way, systematic, teach- and learnable. This precedes any specification of content or method, but already excludes intuitionistic Verstehen (understanding) and the like. But when exactly is a scholarly activity or discourse also empirical? And what does being empirical consist of? Theories abound. A good point of departure is provided by Gebhard Rusch’s Erkenntnis, Wissenschaft, Geschichte (Knowldege, Science, History) (Rusch 1987). His approach is constructivist and operational. Empirical for him are those events, objects, situations or processes which are accessible for manipulation, control, measurement or observation in the observer’s here and now, that is, in real time, and the activities brought to bear on them. In other words, empirical is precisely that which can be directly experienced and/or manipulated by the observer. This sounds plausible on first encounter, but does not reflect the standard understanding of the term and is very problematic for literary studies. On the one hand, purely conceptual
structures could also count as empirical on this view insofar as we can manipulate them in the here and now, and this is counterintuitive. On the other hand, any study of the past, be it natural history, evolution, cosmology, or literary and cultural history, immediately becomes non empirical, since the processes we are studying are not part of our here and now. What is part of our here and now are only traces or products of past entities or events (dinosaur bones, texts). So maybe we should try something else. I would propose accordingly to shift the notion of being empirical from objects to discourses about objects or theoretical languages. A discourse will be considered empirical or having empirical content if and only if it contains essentially and ineliminably terms and claims which refer to identifiable objects, [that is, individuals (with their properties and relations), groups and domains] states, events, processes, acts, actions and activities locatable or located in actual space-time. A term is empirical if it can refer to one or more such individual facts in the actual space-time world, and a claim is empirical if it contains such terms. A claim is empirically true, or better truth-like (verisimilar) if it contains one or more empirical terms and is corroborated pro tempore relative to some actual space-time domain (Popper 1963). Note that only purely conceptual claims can be completely true or fully validated. There can be claims which are empirical in content but whose empirical truth value cannot be determined or even reliably assessed, e.g., claims of what a certain poet felt on the day he wrote sonnet X, where we have no evidence of any kind for what he felt on that day. We don’t want too many of these obviously, but must recognize that empirical claims may vary as regards the ease or difficulty of trying to validate them. (And there are of course many false empirical claims). As I have already suggested, empirical is much wider than experimental or testable. Many natural sciences, empirical by definition, are non-experimental: natural history, geology and cosmology are good examples (none of them can repeat, vary or manipulate the input data). But they are able nonetheless to provide supporting evidence for their claims. Experimental methods in esl can obviously be applied only to contemporary phenomena, and it is also only here that one can formulate any predictions. The experimental in esl can best be understood as that particular subset of the empirical where it is possible to directly manipulate textual and/or contextual data and then observe and measure the results and their variations in different cases (see Bortolussi & Dixon in this volume). In fact, it is roughly what Rusch (1987) defined as empirical. The degree of empirical content of individual claims or of a cluster of related claims varies greatly according to the area of enquiry one is engaged in, and the kind of questions one asks in each case. It is therefore better to speak of the study of literature with empirical orientation or emphasis (Schwerpunkt), rather than
empirical or not tout court. And one could obviously entertain a methodological meta norm: form your enquiry in such a way as to maximize the number of empirically valid claims contained in your discipline (vs. purely conceptual). An empirically oriented study of literature (which includes of course the experimental) can engage in all the activities in which a mature science can engage according to Bunge (1998): isolate phenomena or aspects, describe and categorize, correlate, explain, predict (e.g., future trends in drama writing, what kind of novel will most appeal to group Go), and control, or feed a theory back into the observed system. For example, claims made and experimentally validated in esl about the factors influencing text reception and comprehension by school children can lead to change in methods of teaching and in curriculum, with a resultant change in reception and comprehension output (Zyngier et al. 2007). Moreover, purely conceptual models, e.g., of narrative as developed by narratologists for example, are indeed not initially empirical, but could be employed in experimental work as hypotheses about actual world phenomena, for example, to test whether or not they possess psychological reality (hence empirical validity) with respect to the literary behavior of given reader groups (Bortolussi & Dixon 2003). And the same applies to models of thematic interpretation (Louwerse & van Peer 2002). Even more than that: because literature processing is a culturally learned competence, the educational system can turn any purely scholarly text model into part of the competence of non professional readers, hence into literary communicative reality, or observable fact of the functioning of the system of literary communication. This is evidenced inter alia by the introduction of structuralist concepts into literary education on the high school level in France.
2. Literary historical study as part of esl Of all areas of the traditional, text-focused study of literature, literary historiography is an ideal place to exercise the nexus between semiotic objects and their actual world contexts, as it starts from texts and text types and then proceeds to Kommunikate. Work here is of necessity observational rather than experimental and simulative, but can be highly empirical in its contents. In fact, any literaryhistorical study of text in context is by definition empirical in its import, since it wants to know about individual facts: not only what is there textually but also the following in the actual world: When, where, how, by whom and for whom was this semiotic object produced? Why is it like this and not otherwise, that is, why this specific kind of semiotic product at this particular time and place and by this individual? How was it transferred over time and space, by whom, and to what end?
How was it received in the sense of meaning ascription, evaluation and functionalisation in different groups (= time + place), and why like this and not otherwise? How, when, where and why was it transformed or adapted? In case of change it would also like to know why this kind of change, why at this time, and how did it actually occur (mechanisms, stages). And similar questions can be raised about text types, genre models, styles, metaphors and so on. Notice that the point of departure is textual: a semiotic object or aspect (a formal or narrative structure, a style, device, genre, text type, fictional world etc.) as defined and identified within and through the rational conceptual discourse of the scholar, and only then correlated with the empirical component of literary communicative activities. One thus starts from the textual and then proceeds to reconstruct the contexts of production, mediation and reception in which it occurs and changes, the network of relations between features and structures in the semiotic object and the situations of the system- participants and the activities undertaken by them. The poetics and explanations given by the participants themselves are part of the data used by the scholar in formulating his own how and why theories about the very same phenomena, and serve as the object of scholarly historical explanatory hypotheses. Here are a few examples of key questions as regards production formulated by Viktor Zhirmunskii (1996). In literary historical work, he says, one is interested in exploring the author’s artistic path, with the work being understood as a dynamic, changing entity (Utterance, Kommunikat) and not an isolated closed object. One is thus interested in the author’s laboratory: the various stages from first draft to finished product and the factors behind these modifications, in a word, in the Entstehungsgeschichte (history of origin) of this product. One is equally interested in changes the author himself introduced in a work of his in successive editions, and the reasons for these changes as reflected say in diaries or comments to others. Equally interesting would be to find out why an author decides to include or exclude certain works from his collected works. And as for reception, it is of vital importance for our understanding of the literary historical process that we gather as much as possible of the original audience’s reactions (reviews, comments, debates) to a work when it was published. For this audience, the work is not yet a monument, but rather a current event, a communicative act by an author that is performed in a particular situation, or an utterance that is the object of immediate reaction. We would like to know how the work as Kommunikat inserted itself into the contemporary cultural situation and what role it played in space-time and group specifiable ongoing debates and conflicts. Having noted the numerous empirical aspects of literary historiography with respect to text production, mediation and post processing, I would now like to proceed to three further methodological observations.
1. The incompleteness of the purely conceptual model of the literary historical process is evident as soon as texts or text types are described as being influenced by other texts or giving rise to them or mediating between some A and B. Surely semiotic objects can do none of these things. It is only a human agent who reads text A, processes it, forms an interpretation or representation of it in his mind and then proceeds to combine it with some other texts and text models stored in his memory in order to produce text B. Intertextuality on the level of individual texts, genres or discourses is not an abstract disembodied semiotic process but always a product of actual reception processes. Moreover, being intertextual with respect to a given earlier text and being so in a particular manner are both mere options for the author of a newer text. The author, as an agent, has certain goals in mind, which, he judges, this intertextual exercise will enhance. esl will have to reconstruct these goals and reasoning process and seek to explain why in this historical situation exercising this intertextual option was considered an effective or optimal solution (Margolin 1999). Much the same holds for general “literary change”, which Titzmann (1991: 395–398) describes as follows: Change from A to B in the literary system means the following: (1) Up to T1 a group of people G produced texts out of which the scholar can abstract a descriptive poetic system S1, and this system was dominant. (2) Beginning with some later point T2, a group of people created texts out of which we can abstract S2. S2#S1, and was dominant at a later time section. Even if literary change can be explained inside a scholarly constructed poetics as an attempt to solve a problem inherent in the current state of the system, the choice of one particular solution out of the myriad possible ones can be explained only by reconstructing the actual reasoning of the system participants through the meticulous and comprehensive collection of their various utterances on the subject and an analysis of these very utterances and their inner logic. The absence in many cases of any records of individual or collective reaction to and reasoning about literature by system participants sets rigid limits to empirical historical work, but does not obviate its significance. And the same goes for the reconstruction of an author’s or a group’s horizon of production, to coin a phrase for a well- known cultural and social context. 2. The picture presented so far of the poetic-empirical relation is however only one half of the story. In the study of much of pre-print literature, the order of exploration is actually the reverse of that concerning newer literatures, and another stage is added. History of transmission (Überlieferungsgeschichte), for example, is a historical sub-discipline much developed in the 19th century for the study of ancient and medieval texts. The physical vagaries of texts, including palimpsests and copies and their dating; alternate versions, redaction, recension, filiation, presumed translations, establishing the standard version of a text, its Urform or
a text-type invariant are all operations based on factual historical reconstruction, and so are issues of attribution, such as who could have written a given anonymous text, or at least in what milieu, hence where and when. And the same goes for examining whether a text attributed traditionally to a particular author could probably or possibly have been written by him or not. In much of medieval studies scholars have treated the answering of such questions as the only valid literary-historiographical activity, and suspected anything beyond as subjective or anachronistic. Coming out of a positivistic nineteenth-century understanding of the nature of scientific discourse, they were ready to reconstruct the con-text, but unwilling to formulate a poetics of the text in its midst. The result is that since these facts are not functionalized and made to illuminate textual features and structures, the contemporary student of literature finds them devoid of literary significance. But things may have by now gone too far in the opposite direction. Taking for example the recent Pleiade one volume edition of the various Tristans, students, and non specialist teachers of literature as well, tend to read these texts as finished semiotic objects, ready for purely conceptual poetic description, theorizing, comparison and interpretation, forgetting that the very form, chronology, filiation and multilingual historical relations of these texts as presented in the Pleiade edition are not self evident and independent but rather the product of a long and often hypothetical process of reconstruction. In other words, in this area the elementary givens of scholarly poetics, i.e., the individual texts themselves, are end results of empirical work, often only hypothetical constructs, and not simple initial givens as in the modern age. There would have been no possibility of intrinsic poetic analysis of any kind of much pre-print literature if the empirical philologische Kleinarbeit (philological legwork) of establishing texts and dating them had not been undertaken initially. As a second stage one could obviously seek to relate these texts to surrounding forms of thought and social forms; but this will now be the second empirical twist of the literary-historical tale. The full order will hence be: empirical (textological and historical); elaboration of scholars’ own poetological descriptions of these texts; relating textual patterns thus defined to a wider actual cultural and social context. But even this is too simplistic a description of the relations between the empirical so called facts and the scholarly poetic constructs. In deciding how to determine the standard or best version of a medieval work, for example, and in considering matters of relative dating among versions (Stage 1) – both seemingly straightforward empirical matters – poetological, theoretical and even value considerations can play a major role. One example would be the assumption that texts require closure, and that the version with the clearer one is therefore to be preferred for canonizing, anthologizing etc.
3. Finally, in the context of empirical historical work, reception and post processing are inevitably collapsed together. We have no access to past receptional activities except as recorded in texts, which ultimately makes textual analysis the only major method of verifying any claims made in the historical empirical study of literature. This also ensures that until the age of current opinion polls, mass questionnaires and the like, most post processing texts available to us will originate with professional participants in the system of literary communication (Booth’s professional readers [1988]) and not from the listening or reading public as a whole, with some notable exceptions of recorded reactions by nobility, leading public figures etc. This is of course a major limitation for the comprehensiveness of any description, with the one comforting factor that the professional agents within the literary system usually play the major role in determining its course. But this too is an empirical claim which, if it cannot be verified, will inevitably have to be assigned the status of plausible, but unproven, and in some cases unprovable, hypothesis about literature as an actual historical communication system. Theory formation and confirmation are bound to meet sooner or later in the discussion of any enquiry possessing empirical content. Now that they did, I would like to make some general remarks on the ways we can support claims made in esl.
3. Methods of verification Any empirical proposition occurring in the study of literature should in principle be open to verification (in the sense of assessing or testing its truth claim) through one or more scholarly acceptable and shared methods or procedures, which, once again, cannot be limited to the experimental. As we have just seen, the vast array of the activities concerned with the reconstruction of a literary historical context involves the critical analysis (Auswertung) of relevant texts, serving as the data store, and the formulation on the basis of this analysis of descriptive-correlative or explanatory (causal, teleological) claims concerning the literary system. Texts are our only source of information here, and our only basis for formulating scholarly claims as well. For clearly we have no direct perceptual experiential link to the past and the past actions or activities of any human agent or group, and certainly not to the minds of participants in the system of literary communication. As in much other historical study, the assessment of the validity of these scholarly claims is itself also restricted to textual/symbolic activity. For example, checking whether a set of initial scholarly claims, based
on the analysis of a certain non-literary text corpus, indeed constitutes adequate evidence (basis, support) for formulating a given scholarly claim about the literary text or the nexus between it and its context (=all New Historicist claims, specific or general). In this process of assessment, the initial data, both literary and non literary, consist of utterances made in determined space time locations by actual participants in the system of literary communication, and the critical assessment of any scholarly claims about these data employs both inter-subjective canons of argumentation and inferencing (“rules of evidence”) shared within a scholarly community, and background theories about human action, especially a symbolic one. The assessment procedure is hence both rational and empirically-based, but neither directly observational nor experimental. Direct observation, as well as experimentation, is feasible only with regard to what is currently and directly accessible, such as measurable (quantifiable) properties of texts on one hand and, on the other, the behavioral acts, including of course verbal ones, of contemporary individual and group participants in one or more of the four components of the system of literary communication. Research (testing, verification) methods here include the whole array of social science and psychology (especially cognitive, discourse processing, developmental) methods. With respect to the text-reception activities of individuals or small groups one can of course employ experiments of various kinds, with the setting being controlled and the input data being subject to manipulation by the scholar. But as we know, even here matters are complex, since the interpretation of experimental findings is highly theory-dependent. This, however, is the same in the natural experimental sciences as well. Finally, and especially in the areas of story generation and recognition, more and more scholars have been resorting to computer simulation. Here one writes a computer programme which will be able to generate certain (types) of stories given a set of formation and transformation rules, or to draw conclusions (“interpretation”) from a simple story used as input data, based once again on a set of inference rules. The idea, as in much other work in cognitive science, is of course to simulate that which we can never directly observe: the mental operations involved in story production and comprehension. Any unacceptability or inadequacy, respectively, of resultant stories or inferences with respect to what actual readers do is understood as indicator of the additional or different capabilities deployed in the corresponding human operations. Conversely, acceptability/adequacy of a programme are interpreted as evidence that the programme may possess psychological reality, in the sense of accounting for what actually happens in our minds when we produce or interpret stories, but we can never prove that it does.
4. The study of literature as itself an object of historical-empirical study Since the 1960s it has been widely accepted that any scientific discourse, regardless of its specific content, is a product of a particular cultural-historical context and that it is one discursive activity among the numerous ones in existence at any given point in time. More attention has also been paid to science production as a social institution, and now there is even a sub-discipline called sociology of science. If so, then the particular nature and theoretical contents of the study of literature at any given point in time can legitimately become the object of precisely the same kind of empirical historiographic study to which esl submits the literary system. This will involve embedding or anchoring the study of literature and its contents – especially literary poetics and general theories – as they exist at any period within the wider contemporary scholarly discursive context (philosophy, linguistics, political thought) and asking for example why particular models of narrative or even the whole discipline of narratology came into being at this particular time and place, why they had the specific contents they did, why did they change or get widely accepted or wane away at a given time, and how precisely did all this come about. In such a study, one would also have to take into account that science is an activity carried out by actual human beings who enter into complex exchanges with one another, that science is a social institution, and that personal and institutional factors may play a key role in shaping the theoretical contents of a discipline and effecting changes in it. Scholars’ educational background could also be called upon to explain why they formulated certain theories by way of continuation or rebellion. Such a historically contextualized approach to the dynamics of scientific activity may even reveal a great similarity between the regularities or life cycles of literary and scholarly schools (See for example Seyffert 1985). In the case of Russian Formalism for example one can show that their claims about the dynamics of literary change apply to the dynamics of their own theoretical transformations, a thing only a retrospective meta level study can reveal. And of course general historiography has been the subject of numerous contextualising meta-historical studies in the last 30 years. But this does not lead to a relativist or anarchic situation as regards the validity of any literary theory. Contextualising and describing (or explaining) how and why a particular poetic theory emerged in a particular time, place, and manner and why it is thus and so does not in any way reflect on its logical validity or on the scope of its applicability to actual literary phenomena. And from the fact that a literary theory is formulated at a specific time, place and context it does not follow that it will be considered valid only in that frame or that its validity is restricted to literary products of that time frame. This pernicious and widespread relativist non sequitur needs to be exposed and corrected. And this leads me to my final section: some urgent tasks of the esl as I have described and discussed it.
5. Some theoretical and meta theoretical tasks for esl esl as an experimental activity is dependent on methods developed in psychology and in the social and cognitive sciences (discourse processing). One pressing task here is to test on individuals or different groups the psychological and cultural reality of numerous theoretical text models, e.g., narratological ones (do readers really perceive each case of free indirect discourse (see Hall in this volume)? Do they make great mimetic capital out of it? Or of the first vs. third person narration? Do they distinguish sharply between fictional and non fictional texts as regards truth expectations the way Dolezel’s (1999) theory does? In what stages do they build a text world (Werth 1999)? On what basis do readers decide that a narrator is partially or wholly unreliable? Poetological: Is parallelism so perceivable and prominent in the reading of poetry and do readers use it as a guideline for interpretation? Dramaturgic: Does the narrativization of contemporary drama, described in detail by Pfister (1988) and Nünning (2002) among others, strike them as a basic departure, as something highly perceivable, as illusionszerstörend (illusion-breaking) (Wolf 1993)? Answers must then be relativized to cultural, educational and other social factors. It would also be most valuable to try and operationalize and then test global armchair theories of narrative sense-making like Iser’s (1978). Does Iser’s influential theory have psychological reality, or is it an introspective description of what he thinks he himself and other professionals are doing when they make sense of a text as narrative? Or is it a set of suggested norms of sense making, based on abstract general textual and psychological theories? Either way, it would be nice to find out. Within a historically-oriented esl, it would be useful and informative to look for small to middle range, case based regularities, instead of putting forth untestable or trivial universal claims. Examples of such informative empirical studies are van Peer’s (2001; 2008) proof that it is not always the conformist who is preserved (Romeo and Juliet), or the correlation he has established between the rise of new media and of new text types (writing, press, electronic). Intertextuality as defined and described by Hempfer (1987) for example needs now to be explored as complex practice and as process on the basis of case studies, the way he himself has been doing with Ariosto. Instead, we currently have either cumbersome morphological classifications or wild post-structuralist claims, both of which have led to the abandonment of this enormously important aspect of literary dynamics. On the meta theoretical level, one should reread RS Crane’s (1967) Idea of the Humanities and appropriate its myriad insights, mostly case based, how to reason and theorize in an empirically valid (supportable) way about matters of literary diachrony. This aspect-by-aspect way of proceeding is much more useful than
current speculations on the handling of time in general and on the conditions of possibility of any historical enquiry. Also on the meta theoretical level, it would be interesting to see in what ways 19th century positivist literary historiography (context and low level facts only) can be re-functionalised and integrated into the current one. Conversely, one would also need to examine whether such positivism is really theoretically innocent and does not presuppose some poetics of its own. And, finally, New Historicism. Does it provide a methodologically valid historical-empirical paradigm if we look at its practice and ignore some of the accompanying rhetoric and ideology? After all, it does seem to have a solid text-context orientation.
References Booth, W.C. 1988. The Company we Keep: An Ethics of Fiction. Berkeley CA: University of California Press. Bortolussi, M. & Dixon, P. 2003. Psychonarratology. Cambridge: CUP. Bourdieu, P. 1984. Distinction: A Social Critique of the Judgement of Taste. R. Nice. Cambridge MA: Harvard University Press. Bunge, M. 1998. Philosophy of Science. New Brunswick NJ: Transaction Publishers. Crane, R.S. 1967. The Idea of the Humanities. Chicago IL: University of Chicago Press. Dolezel, L. 1999. Fictional and historical narrative: Meeting the postmodernist challenge. In Narratologies: New Perspectives on Narrative Analysis, D. Herman (ed.), 247–273. Columbus OH: Ohio State University Press. Hauptmeier, H. & Schmidt, S.J. 1985. Einführung in die empirische Literaturwissenschaft. Braunschweig: Vieweg. Hempfer, K. 1987. Diskrepante Lektüre. Wiesbaden: F. Steiner. Iser, W. 1978. The Act of Reading: A Theory of Aesthetic Response. Baltimore MD: Johns Hopkins University Press. Louwerse, M. & van Peer, W. (eds) 2002. Interdisciplinary Studies in Thematics. Amsterdam: John Benjamins. Nünning, A. & Sommer, R. 2002. Drama und Narratologie: Die Entwicklung erzähltheoretischer Modelle und Kategorien für die Dramenanalyse. In Erzähltheorie transgenerisch, intermedial, interdisziplinär, V. Nünning & A. Nünning (eds), 105–128. Trier: WVT. Margolin, U. 1999. Formal, semantic and pragmatic aspects of metatextuality. In Comparative Literature Now: Theory and Practice, S. Totosy de Zepetnek et al. (eds), 153–163. Paris: Honore Champion. Pfister, M. 1988. Das Drama: Theorie und Analyse. München: Fink. Popper, K.R. 1963. Conjectures and Refutations: Growth of Scientific Knowledge. London: Routledge and Kegan Paul. Rusch, G. 1987. Erkenntnis, Wissenschaft, Geschichte. Frankfurt: Suhrkamp. Schmidt, S.J. 1980. Grundriss der empirischen Literaturwissenschaft. Braunschweig: Vieweg. Seyffert, P. 1985. Soviet Literary Structuralism. Columbus OH: Slavica.
Titzmann, M. 1991. Skizze einer integrativen Literaturgeschichte und ihres Ortes in einer Systematik der Literaturwissenschaft. In Modelle des literarischen Strukturwandels, M. Titzmann (ed), 395–438. Tuebingen: Max Niemeyer. van Peer, W. & Chatman S. (eds) 2002. New Perspectives on Narrative Perspective. Albany NY: Suny Press. van Peer, W. (ed.). 2008. The Quality of Literature. Studies in the Evaluation of Literary Texts. Amsterdam: John Benjamins. Werth, P. 1999. Text Worlds. London: Longmans. Wolf, W. 1993. Aesthetische Illusion und Illusionsdurchbrechung in der Erzählkunst. Tuebingen: Max Niemeyer. Zhirmunskii, V. 1996. Vvedenie v Literaturovedenie. St. Petersburg: St. Petersburg Univerisity Press. Zyngier et al. 2007. Revisiting literary awareness. In Literature and Stylistics for Language Learners, S. Zyngier & G. Watson (eds), 194–209. New York NY: Palgrave-Macmillan.
Empirical research into the processing of free indirect discourse and the imperative of ecological validity Geoff Hall I think what this piece of research shows is, again, how much more complex the world, including the world of literature, is than we initially surmised. I would argue, however, that it is precisely because the world is so complex, that we cannot do without empirical research. If our understanding of literature is to advance with any degree of success in heightening the reliability of that understanding, then empirical investigations will be unavoidable. (Van Peer 2002: 23) This paper uses two principal examples to argue that experimentalist research paradigms in empirical literary research can be suggestive but that results are not easily extrapolatable to actual real world literary reading events, particularly where, as with free indirect discourse (fid), the phenomenon to be investigated is demonstrably complex, multifaceted and highly contingent. More broadly, the paper raises the issue of whether in fact most literary reading is not typically as complicated as fid, in which case complementary or alternative research approaches may be needed. I close by advocating more nuanced qualitative or ethnographic approaches which respect the complexity of the phenomena under investigation to achieve better understanding, even at the possible expense of seductively neat graphs, tables and statistics. As van Peer suggests in my opening epigraph, empirical research at its best can be highly suggestive. My argument is simply that we need always to remember that what we think of as empirical research should not be limited to experimentalist paradigms. The empirical literary research community will be able to say more useful things about fid and the wider complexities of literary reading by complementing more experimentalist work with more contextually sensitive investigations, to the mutual benefit of both. Keywords: free indirect discourse, enthographic approaches, ecological validity, textoids, empathy.
1. Introduction The strength of experimentalist research is that it can give us clear answers to clear questions. The downside is that this clarity is sometimes achieved at the cost of simplifying the phenomena under investigation so far that doubts arise as to the utility of the findings. In what follows I illustrate those perhaps platitudinous observations with specific reference to some interesting recent published research into readers’ processing of free indirect discourse, but also wish to raise the question of whether the weaknesses I perceive cannot be more widely found across other experimentalist empirical research on literature reading. In one of the most respected accounts of fid to date, McHale (1978) concludes that: ‘the essential character of literature itself is inscribed in miniature within it [fid]’. So far as fid goes, then, the underlying assumption, that there is a clear answer to be found, specifically that readers do in fact assign words and sentences on line, more or less sequentially, and definitively, to notional ‘voices’ of characters or narrators, is, I suggest, a misrepresentation of more complex reading activities, where modes of processing, time points and who is reading what and under what circumstances, mean that fid simply isn’t processed like that in (if I may for now take a problematic term for granted) ‘ordinary reading’. Like literary reading more generally, the very essence of processing of fid is a more nuanced and sensitive, highly contextualized process that can be reduced if not traduced by such representations. This is not at all hostility to empirical research as such. (See accounts of some of the valuable findings and questions raised by empirical research into literary reading in Hall 2005a). But without such sensitivity those who doubt the value or wisdom of empirical approaches will indeed only be further alienated. The issue is one of methodology and of research design, ‘external validity’ (adequacy to the real world) rather than any criticism of procedures and interpretations of a given experiment in itself (‘internal validity’). Ultimately this paper signals the need for methods adequate to the investigation of discourse processing which fully recognize ‘discourse’ as utterances which draw on but also always exceed the linguistic limits of the surface words and sentences of which the discourse is composed. The argument will be pursued primarily by considering two important recent studies of empirical research into processing of fid, not because they are weak, but precisely because their rigor and conscientiousness – even imaginativeness – enable us to see these underlying basic problems of methodological approach all the more clearly, raising urgent questions about what exactly the findings being reported tell us. To anticipate the possible charge of simply being a negative wrecker sniping from the sidelines, I close by arguing my alternative but again in full acknowledgement that this kind of study too can be critiqued for its limitations. A complex field of activity (literary reading) inevitably requires a multiplicity of approaches to deepen our understandings.
Chapter 2. Empirical research into the processing of free indirect discourse
2. Study 1. Sotirova (2006) At the time of writing, Sotirova’s ‘Reader responses to narrative point of view’ is the fifth most widely ‘hit’ article on the website of the journal Poetics. This is fully justified by its sophisticated account of fid as well as the intrinsic interest of the experiments conscientiously reported. Indeed I wish to stress at the outset that my own account of the research here cannot satisfactorily re-present the careful subtleties of the paper which interested readers should consult first hand for themselves. The paper aims to test the more theoretical and speculative accounts of critics who lead thinking on fid against empirical and linguistic evidence of voice, and ordinary readers’ accounts of their readings of selected passages from D.H. Lawrence’s Sons and Lovers. (‘Ordinary readers’ = students and some faculty, as usual in such work). The popularity of the article also suggests, of course, that fid is indeed for many researchers in the literary field, a key concern, as indicated in my introductory section. Sotirova rightly quotes van Peer (2001: 337) in her introduction: The theoretical models that we develop also have their limitations, and may contain mistakes. Insofar as these models are based on armchair analyses of texts, a confrontation with concrete readers’ reactions may tell us where precisely the incompleteness of our theories and the mistakes in our models are to be found.
Sotirova sets out then to test empirically what she calls the ‘single’ and the ‘dual voice’ hypotheses which she distinguishes from her perceptive analytic review of the more theoretical literature. The single voice hypothesis she attributes to theorists such as Banfield (1982) and Fludernik (1993). I should declare here that I have argued myself for a comparable view of fid as ‘writing’ (Hall 2005b, closer to Fludernik than Banfield). For theorists of this first persuasion, there is no need to imagine distinct consciousnesses or even ontological ‘beings’ such as narrators and characters speaking and thinking to account for the forms purportedly ‘reported’ speech and thought take. Rather writing, particularly advanced creative writing, is simply a more flexible and resourceful instrument than is recognized in our rather monological speech-oriented grammars. The dual voice hypothesis, on the other hand, associated with Bakhtin (1981) and Voloshinov (1973), but also Pascal (1977); Toolan (1988); Cohn (1966); Leech & Short (1981) and others, points to empirical stylistic features of fid which can be taxonomically catalogued, suggesting for those theorists two subjectivities or ‘single voices’ co-present, that of the narrator (past tenses, third person references) and that of the character (proximal deictics, expressive language etc.), hence ‘dual voicing’. We should note here in passing that dual voice theories posit the function of fid as to evoke empathy or irony towards a character in the reader. We return precisely to the important question of empathy in discussing Bray (2007) in
section 3. To understand however that empathy and irony are rather crude polarities to try to apply in cases of sophisticated literary creativity, the passage Sotirova herself uses from the second part of Sons and Lovers, and the differing interpretations of the sentences among her readers, sufficiently suggest for me that literary response is not usually that simple or schematic. The passage is deliberately chosen for its complexity (p.112). The impulse to chart and catalogue in order to understand can thus arguably lead to misrepresentations (underrepresentations) as well as to the greater clarity van Peer (2001, quoted above) claims. We return to these difficult issues in the Conclusion to the chapter but I would go so far as to claim that the function of literature is to encourage readers to move beyond the reductive polarities of psychologists and even philosophers. Sotirova herself acknowledges that it can be difficult ‘to identify [fid] on the basis of formal criteria alone’ (p.111), indeed that the frequent difficulty if not impossibility of assigning every utterance to a specific ‘voice’ can be the source of power of fid as a literary device, that it ‘destabilizes’ our sense of such an easily knowable and comprehensible social universe where (as our governors would prefer) blame, credit and responsibility are easily assigned. Literature exists to confound such political and legal oversimplifications. Life is rarely if ever that simple. I would argue that we need to move beyond the residual quantitative urge of ‘dual’ voicing, with its implicit metaphor of distinguishable voices, to discussions of language as discourse, a Bakhtinian understanding of the multilayered polyvocal complexity of any use of language, more accurate in its final indeterminacy (language as a site of social struggle). Returning more directly to Sotirova, there are obviously real and important differences between the models of Banfield and Fludernik, or (say) Short and Bakhtin, and Sotirova (e.g., p.110) is the first to admit that empirical findings to date as well as theories are often contradictory and so suggest a complex object of investigation. But the broad distinction between theories made here is well argued and provocative (‘single’ vs. ‘dual’). Once it is established, Sotirova turns to test the two models empirically, and the adequacy of the empirical work offered is the central concern of this section of my chapter, against, of course, a baseline concern for ecological validity (‘Life’ above; see also Conclusion). Sotirova presented 86 informants in four separate sessions with an extract from Sons and Lovers to investigate whether they seem to respond to ‘dual voicing’ and also to investigate apparent sensitivity to formal features in the writing of the kinds mentioned above, including pronoun shifting, tense deictics, clause structure, lexical features, and explicit verbs of consciousness or communication. Questions were designed to prompt ‘readers to attribute the viewpoint in a narrative passage to one or more of the narrative personae’ (p.112). The humanization of the interpretative reading process is to be noted. Second, questions were designed to
Chapter 2. Empirical research into the processing of free indirect discourse
chart reactions to a complex passage ‘sentence by sentence’. This is not necessarily to say that interpretations will develop in that way. The relevant unit could be the clause or smaller, or the paragraph or larger, but Sotirova argues that in this way the developing interpretations of her readers as well as individual reader variation can be empirically studied. Again, we must note that what is therefore being studied is a particular reading process or occasion of reading, under particular circumstances (being prompted to respond sentence by sentence). How far might other readings depart from this prompted style of reading? How far would that matter for the validity of any conclusions reached on the basis of this study? Sotirova’s working hypothesis, for which she seeks evidence, is that readers of the passage will be attempting to distinguish “Miriam’s” words from “Paul’s” from the “narrator’s”, three in principle distinguishable voices or characters. The paragraph is then presented to readers with sentences numbered, again prompting a sentence by sentence approach that is unlikely in a more naturalistic reading of a novel. A single voice hypothesis of course would find these proposed distinctions suspect, preferring instead to look for evidence that a reader seeks understanding of ‘what a writer is saying’ which goes beyond (and not necessarily even ‘through’) such a logocentric dynamics of voice (characters, narrator), moving more directly to process the linguistics of the system of writing offered (compare arguments that readers respond to ‘textual anomalies’ in Ron 1981). Sotirova provides her respondents with a ‘Don’t know’ option, but of course educated test takers will recognize fully that this is a dispreferred response, likely to be unhelpful to a researcher and to be avoided if possible. She finds, in any case, that all four of her groups tend to look beyond the level of the sentence even as they respond to the sentence level probes. But even if these ‘subjects’ tried to respond holistically to the wider paragraph, another relevant factor for ‘real world’ readings, surely, is the use of ‘textoids’ in research experiments, as Graesser & Kreuz (1993) called them, decontextualized extracts designed to test an experimentalist’s hypothesis rather than to simulate something like authentic conditions of reading, where ethnographic research would be more interested in the participant perspective. Even if it is accepted that sentence by sentence interpretations occur in ordinary whole novel reading (surely life is too short?) those interpretations would be informed by a developing or cumulative understanding of the ‘text world’ of the novel, including our ‘mental models’ of the characters (Werth 1999; Emmot 1997; Gerrig 1993). Arguably, the readings of those who recognized the author’s style (three did, p.115) or even the precise novel (another three did) are of more interest as closer to a real world possible experience, despite these imposed conditions of reading. Bortolussi & Dixon (2003), for example, have established empirically that repeated readings make a real difference to interpretations. I certainly noticed
many features of Sons and Lovers for the first time when I found myself teaching it recently even though I had read it many times before. I ‘heard’ much more of Narrator and less of Paul than I had as a younger male reader, but again finally found these ‘voices’ an inadequate response to the subtleties of Lawrence’s writing. One definition of literature admittedly is that it is text that receives unusually close linguistic scrutiny (compare Zwaan 1993) and which may be read repeatedly with, if anything, growing attention to verbal details and changing understandings. But also we know that the reader who reads is not a fixed variable either, so that different details will be noticed differently in successive readings even by the same reader. My ecological doubt here is the kind of attention Sotirova’s readers are prompted to pay to the researcher’s textoid, as opposed to more naturalistic novel reading activities which, it is proposed, this research can tell us more about. In the light of such reservations, Sotirova’s move is interesting. One reading is that Sotirova was finding instability of meaning and attribution, possible multiple ‘speakers’, as allowed for in any classic account of fid (so that ‘dual’ comes to mean ‘two or three’ – or more?). The more radical possibility (‘single voice hypothesis’) is that the limitations if not final impossibility of character/ narrator assignment as a strategy are becoming evident. Remember, after all, that ‘characters’ are literally marks on a page. Perhaps readers are savvy enough to recognize that. Arguably, the more consciously they are asked to process these marks the more futile the attempt to assign supposed utterances in writing to non-existent speakers will seem. It is worth noting in this respect Sotirova’s own reference to Lodge (1990: 66–7; cited Sotirova p.115) which refers to the Bakhtinian ‘stylization’ with which Lawrence frequently represents discourses: nobody, after all, is ever supposed to have spoken exactly ‘that’, for Lawrence or for his more attentive readers, even in his earliest more realist fiction. We must remind ourselves that Lawrence (or any other writer) is never ‘reporting’ actual words; he is writing them. We need at least a more sophisticated understanding of ‘reporting’ if we are to persist in using the term (Hall 2005b; the parallel with newspaper ‘reporting’ is close). The final important aspect of this study to note is the tendency Sotirova found for those more experienced literature readers or with more experience of discourse analysis (staff and research students in the English department of a U.K. university as opposed to undergraduates) to report more blending of voices, in particular, (surely not coincidentally in line with much published professional criticism) a blurring of ‘voices’ of Paul and the narrator in Part 2 of Sons and Lovers. This is congruent with many other studies of reading and of the reading of literature,
. In the light of her own first findings – in a revised version of her questionnaire to ask for ‘one or more’ and then ‘more than one’ possible speaker of a sentence.
Chapter 2. Empirical research into the processing of free indirect discourse
which seem to show that ‘expert’ readers are better both at noticing detail and at developing more coherent synoptic interpretations. (Some key studies reported in Hall 2005a). A ‘single voice’ hypothesis might see this development of literary reading expertise as a recognition of what I have called the linguistics of writing as opposed to the logocentrism of humanist ‘voices’, or in more traditional terms the development of ‘toleration of ambiguity’, if not of apparent contradiction, as a key lesson the reader of literature learns. Here I might differ from a conclusion Sotirova draws that her more experienced readers notice more ‘dual voicing’ and less experienced readers read more monologically (p.119). Another way to express this which seems to me more faithfully to report her findings is that more experienced readers ‘hear’ more possibilities and do not close down interpretations as readily. ‘Dual voicing’ is too reductive a notion and too misleading a label to express the openness we are shown through this research. The voicing perceived (if that is what it is) is multiple, indeed often quantitatively undecidable. The findings exceed the experimentalist’s model. It is the value of this piece of research, but also a source of unease with it. Although there are undoubted tendencies toward majority interpretations of a given sentence in Sotirova’s data, she honestly reports also the extensive variability of interpretations, that nearly every sentence was actually attributed to every supposed ‘speaker’ or combination of ‘voices’ by one respondent or another! (Table 3, p.117) This to me is again suggestive of the need to think beyond the attribution game. In fact, in Sotirova’s own words, ‘there is no clear-cut boundary between narration and free indirect style’ (p.119; compare conclusions of Short, Semino (1997) and others working on the Lancaster University sptw project; Semino & Short (2004)). How far then are even Sotirova’s ambivalent findings supporting the relevance of some idea of ‘dual voicing’ the consequences or product of a research methodology that asks (to reduce rather crudely) respondents whether they hear ‘two voices’ in an uncontextualised extract, prompts them with the names of these voices and then finds that this represents exactly what they do? Experimentalism typically features ‘questionnaires containing items formulated in the researcher’s terms’ (Schreier 2002: 40), as opposed to participant perspectives; favors yes-no answers as opposed to more emergent models of comprehension. We can assign voices to characters on a sentence by sentence basis if we are asked to do that (at least to some degree, according to this research, though it can be problematic). But is that what readers naturally do? Or better, to avoid difficult constructs like ‘real/ ordinary readers’ and ‘natural reading’, under what conditions of reading do readers do that? Is it, for example, (my own hypothesis) a common strategy of less committed readers to simplify processing to get through all those damned pages of a classic novel? Superficial processing if you will, so that one of the objections to Lawrence,
an aspect of his ‘difficulty’ is that his fiction tends to resist such simplistic assignations; it simply will not be an effective strategy for rewarding reading of his work. Remember, after all, the warning that Lawrence memorably expressed to Garnett, his editor, that ‘you mustn’t look in my novel for the old stable ego of the character’ of individual consciousnesses (‘characters’ and ‘character’ v. ‘narrator’ divisions), recognizing, like his near contemporary Bakhtin, the multiple interconnectedness of lives and the central role of discourse in the functioning of consciousnesses (Lawrence (1979a) letter of 5 June 1914). At the least there is risk of a kind of observer’s paradox here, so that the observed object takes the forms it does as a result of being observed (compare Hunt 1996 in my Conclusion.) If those questions are too fanciful or ‘theoretical’, what would have been the results of a methodology that tapped into an arguably more typical literary reading in which the ‘ordinary’/ naive novel reader tries to appreciate the (singular) writer’s understanding of reality, the point of view of another (one other)? Would we not have been likely to find then a more ‘single voice’ working hypothesis in the minds of readers? (This may relate to the issue of postmodern expert readers in Sotirova’s study perceiving multiple, even contradictory synchronous voices). For all the intrinsic thought-provoking interest of the research reported (compare van Peer), we seem ultimately to be left with the ‘it’s-all-very-complicated’ understanding we had when we set out. Or ‘single voice and dual voice hypotheses’ both have something to be said for them in terms of empirical evidence, but both seem to over-simplify. The conclusion is inconclusive; fid – a type for literature itself – is by definition indeterminate; it is ‘discourse’ and needs to be studied as such.
3. Study 2. Bray 2007 While Bray’s study is not Sotirova’s, as should become evident in this short account, it bears sufficient methodological similarities as an example of ‘experimentalism’ in empirical literary research into a related area (fid) that I shall give less detail concerning this second example to avoid repeating earlier points in my argument. Again, however, I would begin by emphasizing the real interest of the research despite the equally real misgivings I express in what follows. In Bray (2007) then the experiment consists of those inevitable university student readers this time being confronted with two versions of two novel extracts in an admirable effort to test critics’ intuitive claims of the ‘empathy’ effects generated by fit (free indirect thought). Issues of reader engagement and affective response, as well as of critical reflectiveness are of course central for many to traditional claims for the value of literature in education. Literature teaches us, according to this argument, both to understand and ‘feel with’ fellow human beings (‘characters’, here through fis, Free
Chapter 2. Empirical research into the processing of free indirect discourse
Indirect Speech), as well as to maintain and even develop wider moral standards, if not actually to judge others (with ‘narrators’, here through (F)IT). A key issue is being investigated. And even though results tend to confirm conclusions of writers like McHale (1978) on the unpredictability of effects on readers of textual forms, and of the importance of reader variation and reader contribution to meaning construction (Bortolussi & Dixon 2003), Bray’s is an exemplary study in giving some empirical grounds for such assertions, as well as raising new questions to research. (Compare Verdonk (2002) on the value of stylistics even to confirm as knowledge what critics suspect on the basis of their reading experience to be more widely true.) From my own perspective, as I shall argue, the new questions prompted by this study point to the value of ethnographic or qualitative research. Bray’s experiment is thoughtfully designed, always at least half the struggle for such studies. 24 female subjects (14 English literature undergraduates and 10 English language, in group of 6) read an excerpt from Camilla (Burney 1796) and I am Charlotte Simmons (Wolfe 2004) in the orginal wordings or in a modified version in which FIT wordings were re-written by the researcher as IT (Indirect Thought). The passages are well chosen, as results reported confirm, for their relevance to the female undergraduate readers (ages not reported) and to investigate possible effects of English literary historical development from more controlled forms of fit in Camilla (narrator voice distinguishable or where not extricable, likely to be perceived as present) to Charlotte Simmons (fit more standard in the English language novel by this time, greater freedom from narratorial control). Both passages feature heroines in a moral quandary, but while Camilla agonizes over the implications of taking financial assistance from a man, new college student Charlotte seems to have more contemporary concerns over how she has been seen to behave in public with a man at a party (a move from guilt culture to shame culture?): “How could she have let him keep touching her that way?” (CS quoted Bray pp. 9–10). Readers of all passages were asked to rate ‘How close you felt’ to the heroine on a scale of 1 to 10, where 10 is virtual identification and 1 signifies lack of sympathy with all the fuss- a prompt, that is, to read in a certain way. (The implicit assumption is that both excerpts are competent, even good writing; poorly executed attempts at fit would cloud the waters.) Second, readers were asked to ‘Comment on features of language of the passage that led to your ranking’. Of course this second task could have led to revisions of rankings in the first, as would be normal in literary discussion, even in the 10 minutes allowed, but we are not told if evidence of such revisions was observed. The move from sentence response (Sotirova) to a paragraph is praiseworthy, but we must of course note again, however comprehensible in terms of practicality for this kind of empirical research, reading an extract may not represent well how the complete novel from which the extract is taken would be read, particularly if the blurb had been scanned, the
book selected for herself by the individual reader or recommended by a friend, then read alone at home with a packet of crisps to hand on the sofa and a good music track playing, occasional bleeps as texts arrive on the mobile phone, and so on – in short what I have gestured towards as more ‘natural’ reading conditions. Further, following a standard experimental technique, half the texts these readers read were hypothetical rather than the real thing, though the intention of this is better to understand real readings of real text in this indirect way. Bray’s predictions for his experiment, in line with the speculative critical writing on this subject were that: • •
there would be less empathy elicited by it than for fit passages, less involving ‘affect’-effects responses would be likely to differ more between the two versions of cs than between the two versions of Camilla, because the formal differences are more clear cut for the two versions of cs Charlotte would be easier for these readers to empathize with because the fit is freer, less narrator present, and because her situation is more immediately recognizable to a fellow sophomore
The first and third hypotheses were confirmed by the research, the second disconfirmed (a post-processing effect perhaps). As Bray discusses his results, however, and in particular the responses to his second question (on linguistic features influencing response) we should note that the paper comes most alive – comments rather than numbers. In the author’s own words: ‘The subjects’ comments in response to the second part of the task turned out to be as, if not more, revealing than their rankings’ (p.16) The language of Camilla’s experience had clearly been experienced as too formal for emotion for these modern relatively unschooled literary readers, a barrier to empathy, where Charlotte more closely ‘speaks their own language’. The details of this discussion are fascinating and ample evidence for me of the value of more ethnographic investigations, suggesting, for example, the discursive nature of emotional experience in modernity, and the relevance of literary education to the articulation and development of such discourse in the individual (uses of ‘empathy’ or ‘arouses pity’). This may be the point to remind ourselves that these ‘subjects’ of the experiment are actually human beings (qualitative perspectives, individual variation) as well as being, as human beings, ‘subjects’ in the sense that they are positioned by language and ideology. At the least, as Bray himself importantly concludes, the study shows clearly that ‘despite the instruction to focus on ‘features of language’, the amount of empathy that subjects felt for the two characters often seemed to depend, from the evidence of their comments, on non-linguistic factors’ (p.18).
Chapter 2. Empirical research into the processing of free indirect discourse
4. Conclusion: Experimentalism and the imperative of ecological validity In my title and elsewhere in this essay in passing I have referred to the issue of ‘ecological validity’. Ecological validity is not quite the same as ‘external validity’, the generalizability of a study, as my opening remarks may have too simply suggested. A study might be ecologically valid without being externally valid (because of atypical participants or other features of the situation), while it is difficult to see how a study that is not ecologically valid can be very usefully generalized to wider populations and a greater range of reading events (or how we would know if it could be). Thus ecological validity is the very basic demand that a study actually tells us about the phenomenon it purports to tell the researcher and the readers of that research about and not about a suggestive but frustratingly parallel research universe. A key issue in empirical research of fid, from Banfield on to the studies considered here, has been, I have argued, an ecologically suspect concentration on sentence processing at the expense of discourse processing, or use of decontextualized ‘textoid’ extracts from longer works, or even readings of theoretically possible texts that never were, constructed for experimental purposes, all read under atypical conditions. It is here that experimentalist approaches to date have fallen short of the demands of ecological validity. The key ‘ecological’ concerns (compare Zwaan 1993: Ch. 1) are that sentence by sentence processing may not be how literary reading works, and second, that sentences read out of context will arguably be processed differently than if they were encountered in natural continuous reading events by the same readers. A final question is whether reading of texts that never were (outside the lab.) is not likely to mislead unless results are very cautiously interpreted. As soon as we begin attending to reading, it stops, or becomes something else. This seemed even more true of ‘literary reading’. Our attempts to measure it seemed to cause it to evaporate. Hunt 1996: n.p. Consider next the condition in which an individual is reading pointless, incoherent, experimenter-generated text for no particular purpose. Unfortunately, this has been the typical state of affairs for the majority of the published experimental studies during the last twenty years. (Graesser & Kreuz 1993: 156)
Hunt stresses that the ‘purpose’ needs to be the reader’s own. Again, all readings – experimental or non-experimental – are always inescapably contextual. Thus previous readings or knowledge of an author and perhaps even of reception history, for example, will undoubtedly affect retrospectively later re-readings in terms of what a reader notices and how what is noticed is interpreted (compare my comments on Sotirova’s Lawrence paper). Bortolussi & Dixon (2003)
argue convincingly that individual reader variation can still be explained by larger models of discourse processing but the studies reviewed here, as well as Bortolussi and Dixon’s own work, continue to disappoint in this respect. Empirical studies of literary reading (rather we should say of ‘literary reading related behaviors) too often tend to be treated as an unproblematic window on to real reading events rather than the highly mediated and guided affairs they are. (I of course recognize that the ideal of a pure real reading event is itself a myth given the centrality of education to literature reading practices, but this only amounts to a further argument for more specific and ethnographic literature reading research: compare Hall 2006). fid is a paradigm example to my mind of the trained nature of literary perception. Empirical research of all kinds has indeed tended to confirm this belief. Naive readers (also second language readers) typically simply do not perceive, never mind judge and weigh, the multiple simultaneous even contradicting voices that haunt literature almost by definition for those of us who have invested much of our lives into literature reading and teaching (See Goh 1991 or Takahashi & Roitblat 1994 on second language readers; also interesting Discussion of results in Zyngier, Peer & Hakemulder 2007). We train such readers till they achieve these perceptions or fail our courses. Interpretation itself is a key object for empirical study as some of the studies instanced here have rightly recognized, but, like fid itself, it is (they are?) infinitely complex activities. Semino & Short (2004) and others who have tried to taxonomize fid have always admitted that the phenomena of literary reading themselves ultimately escape our best attempts to net them. Indeed, Free Indirect Discourse is the best label for the issue examined here because it avoids the misleading logocentrics of free indirect ‘speech’ of ‘characters’ in favor of recognition of the written nature of this experience through which we are helped to construct our understandings of the world and our places in it. In concluding we can do no better, in reflecting on Sotirova, Bray and the others, – as often, – than return to the words of the always provocative, reliably stimulating subject of this Festschrift himself, who has contributed so much to the field in so many ways, to underline a fundamental agreement on the need for empirical research beyond issues of what forms ‘empirical research’ needs to take: I think what this piece of research shows is, again, how much more complex the world, including the world of literature, is than we initially surmised. I would argue, however, that it is precisely because the world is so complex, that we cannot do without empirical research. If our understanding of literature is to advance with any degree of success in heightening the reliability of that understanding, then empirical investigations will be unavoidable. (Van Peer 2002: 23)
Chapter 2. Empirical research into the processing of free indirect discourse
References Bakhtin, M.M. 1981. Discourse in the novel. In The Dialogic Imagination, M. Holquist (ed.), C. Emerson & M. Holquist (transl.) Austin TX: University of Texas Press. Banfield. A. 1982 Unspeakable Sentences. London: Routledge and Kegan Paul. Bortolussi, M. & Dixon, P. 2003. Psychonarratology. Foundations for the Empirical Study of Literary Response. Cambridge: CUP. Bray. J. 2007. The effects of free indirect discourse: Empathy revisited. In Contemporary Stylistics, M. Lambrou & P. Stockwell (eds), 56–68. London: Continuum. Burney, F. 1796/1972. Camilla; or, A Picture of Youth. E.A. Bloom & L.D. Bloom (eds), Oxford: OUP. Cohn, D. 1966. Narrated monologue. Definition of a fictional style. Comparative Literature 18: 97–112. Emmot, C. 1997. Narrative Comprehension. Oxford: Clarendon Press. Fludernik, M. 1993. The Fictions of Language and the Languages of Fiction: The Linguistic Representation of Speech and Consciousness London: Routledge. Gerrig, R.J. 1993. Experiencing Narrative Worlds. On the Psychological Activities of Reading. New Haven CT: Yale University Press. Goh, S.T. 1991. Higher order reading comprehension skills in literature learning and teaching at the lower secondary school level in Singapore. relc Journal 22(2): 29–43. Graesser, A.C. & Kreuz, R.J. 1993. A theory of inference generation during comprehension. Discourse Processes 16: 145–60. Hall, G. 2005a. Literature in Language Education. Basingstoke: Palgrave Macmillan. Hall, G. 2005b. Who said that? Who wrote that? Reporting, Representation, and the Linguistics of Writing. In The Writer’s Craft. The Culture’s Technology, M.Toolan & C.R. CaldasCoulthard (eds), 151– 165. Amsterdam: John Benjamins. Hall, G. 2006. ‘Literature as social practice’. In The Art of English. Literary Creativity, S. Goodman & K. O’Halloran (eds). Basingstoke: Palgrave Macmillan. Hunt, R.A. 1996. Literacy as dialogic involvement: Methodological implications for the empirical study of literary reading. In Empirical Apporaches to Literature and Aesthetics. Advances in Discourse Processes Vol. 52, R.J. Kreuz & M.S. McNealy (eds). Norwood NJ: Ablex. Lawrence, D.H. 1979a. Letters. Vol. 1. J.T. Boulton (ed.). Cambridge: CUP. Lawrence, D.H. 1979b. Sons and Lovers. H. Baron & C. Baron (eds). Cambridge: CUP. Leech, G. & Short, M.H. 1981. Style in Fiction. London: Longman. Lodge, D. 1990. After Bakhtin. Essays on Fiction and Criticism London: Routledge. McHale, B. 1978. Free indirect discourse: A survey of recent accounts. Poetics and Theory of Literature 3: 249–87. Pascal, R. 1977. The Dual Voice Manchester: Manchester University Press. Ron, M. 1981. Free indirect discourse, mimetic language games and the subject of fiction. Poetics Today 2(2): 17–39. Schreier, M. 2001. Qualitative methods in studying text reception. In D. Schram & G. Steen (eds), The Psychology and Sociology of Literature, 35–56. Amsterdam: John Benjamins. Semino, E. 1997. Language and World Creation in Poems and Other Texts. London: Longman Semino, E. & Short, M.H. 2004. Corpus Stylistics. London: Routledge. Sotirova, V. 2006. Reader responses to narrative point of view. Poetics 34(2): 108–133. Takahashi, S. & Roitblat, H.L. 1994. Comprehension processes of second language indirect requests. Applied Psycholinguistics 15: 475–506.
Geoff Hall Toolan, M. 1988. Narrative. A Critical Linguistic Introduction London: Routledge. van Peer, W. 2001. Justice in perspective. In New Perspectives on Narrative Perspective, W. van Peer & S. Chatman (eds), 325–338. New York NY: State University of New York Press. van Peer, W. 2002. Why we need empirical studies in literature. In Fatos & Ficcões. Estudos Empíricos de Literatura, S. Zyngier et al. (eds), 17–23. Rio de Janeiro: Faculdade de Letras de Universidad Federal do Rio de Janeiro. Verdonk, P. 2002. Stylistics. Oxford Introductions to Language Study. Oxford: OUP. Voloshinov, V. 1973 [1929]. Marxism and the Philosophy of Language. L. Matejka & I.R. Titunik (transl.). New York NY: Seminar Press. Wolfe, T. 2004. I am Charlotte Simmons. London: Vintage. Werth, P. 1999. Text Worlds. Harlow: Longman. Zwaan, R. 1993. Aspects of Literary Comprehension. Amsterdam: John Benjamins. Zyngier, S. van Peer, W. & Hakemulder, J. 2007. Complexity and foregrounding: In the eye of the beholder? Poetics Today 28(4).
Notes towards a New Philology Donald C. Freeman The decline of academic literary study in the United States as it has moved in the direction of “multidisciplinarity” and “theory” is nowhere better illustrated than by “Sokal’s Hoax,” in which an American physicist, Alan Sokal, wrote a parody of “science studies” that was so effective that it fooled the editors of a “cultural studies” journal into publishing it as a straight article. The study of literature would be better served by open, explicit, arguable “real” theory such as cognitive metaphor, an analytical tool that is illustrated in a sample analysis of a famous passage from Shakespeare’s Macbeth. Keywords: cultural studies, science studies, cognitive metaphor, deconstruction, interdisciplinarity, literary theory, philology
Thanks to a growing body of unanswered criticism, a scandal over “science studies” involving one of the most influential journals of “cultural studies” and “theory,” and a stunning recantation by one of its most renowned practitioners, the ascendancy of so-called literary “theory” may be coming to an end. Critics of “literary theory,” however – and I have long been one – must not lapse into one of “theory’s” laziest intellectual habits; we must remember that in any argument for paradigmatic change, mere critique of the status quo is only half the job. For the other half – a start at recuperating literary criticism and scholarship from the damage that “theory” has wrought – we need to develop and institutionalize a program for literary study that both recuperates the insights of an earlier era and focuses on the best of contemporary research. After a brief account of the present state of affairs, I will suggest such a program: what I call the New Philology – a term encompassing such fields of study as stylistics, discourse structure, narratology, contemporary metrics, empirical
. The present essay is a revised version of a talk I gave at the Deseret Language and Linguistics Conference at Brigham Young University, Provo, Utah, February 21, 1997.
Donald C. Freeman
approaches to literature, the European poetics descended from the Prague School and Russian Formalism, the growing body of research in cognitive metaphor, and enough non-specialist knowledge of contemporary linguistics to do work in these fields. With the New Philology as a basis, scholars and critics of literature and the language of literature can begin restoring literary study to the standing that it once enjoyed and continues to deserve. In its contemporary form, “theory” began in my country (the United States) when a French import solidified its presence on these shores. That import was a product of French intellectual jouissance called “deconstruction,” which became coupled to the distinctly non-jouissant Anglo-American lit-crit machine. Deconstruction has eluded all efforts at succinct definition, but we might characterize it as an effort to interrogate existing paradigms of knowledge by dissecting the unstated assumptions, implicit metaphors, etc., of the language in which these paradigms are expressed. Deconstruction marked the genesis of the new “interdisciplinarity,” a variety of scholarship in which the researcher no longer needs to know much about the “inter” discipline. In the case of this Ur-theory – deconstruction – the “inter” discipline was linguistics. However, by declining to acquire at least the rudiments of modern linguistics, Jacques Derrida and his disciples ignored the vast volume of contemporary linguistic research and theory that resulted from the Chomskian revolution in linguistics dating from the mid-1950s, choosing instead to base their theories on the programmatic work of the Swiss linguist Ferdinand de Saussure as represented in his Course in General Linguistics of more than a half-century earlier. Fifty years is a very long time in linguistics. The scope of interdisciplinary “theory” has broadened from the base of deconstruction into the New Historicism, and, more recently, into what has become known as “cultural studies.” In “cultural studies,” knowledge thought to be the province of sociologists and anthropologists – who are professionally qualified to opine on these subjects – was appropriated by literature professors and their students – who are not. We have seen books by English professors on pedophilia, and dissertations by Ph.D. students in U.S. English departments on birthing and thrift shops. The Miltonist Stanley Fish published a book on literary theory and the law (Fish 1989) – not legal themes in literature, but essays purporting to show how “theory” can deconstruct court decisions, legal principles, and the like. In an earlier essay that became the title piece of his controversial book There’s No Such Thing as Free Speech, and It’s a Good Thing, Too (Fish 1994), Fish took up Constitutional law, proposing that the United States abandon Constitutional guarantees of free speech that have been elaborated in . I should say, the excellent Miltonist Stanley Fish, the best Miltonist of our – or perhaps any – generation. As will become apparent in what follows, however, I find less agreeable Fish’s work on theory and on law.
more than two centuries of jurisprudence. Instead, Fish would assign authority over what speech shall be permitted to whatever group currently holds political power. All of this work has made bold and explicit claims for being “interdisciplinary” or “multidisciplinary.” Recently “theory” has ventured into what its practitioners call “science studies.” Philosophers and some social scientists also are involved in this effort, but I will limit myself to literature professors. The aim of this work is to interrogate – to put in question, to problematize – the scientific method, ideas of empirical evidence, scientific objectivity, scientific “laws,” and so forth. The flavor of this work is perhaps best captured in a comment attributed to (and as far as I know not denied by) Andrew Ross, at the time a co-editor of the leading journal of this brand of cultural studies, Social Text. Said he, “I won’t deny that there is a law of gravity. I would nevertheless argue that there are no laws in nature, there are only laws in society. Laws are things that men and women make, and that they can change.” Imagine yourself to be someone who stands outside this work, who has professional training, perhaps primary professional training, in the discipline that is the “inter”-discipline, the discipline that is being joined to “theory.” Further imagine that you believe the “inter”-disciplinary “theorist” who is writing about your field of expertise knows little or nothing about it. What do you do? You can ignore it. That is what linguists did with deconstruction; I do not know a single linguist who believes in it. Nor am I aware of any published critiques of the deconstructionist enterprise by academics whose primary field of scholarly endeavor is theoretical linguistics. Although linguists have revolutionized many literary topics such as metrics, narrative, poetic form, and metaphor, I do not recall more than a halfdozen conference papers or published articles on deconstruction at meetings of the Linguistic Society of America or in that body’s journal, Language. If you can’t ignore “theory,” you can take up arms against it. John Ellis, who has published extensively in both literature and linguistics, wrote an annihilating critique (Ellis 1989) that sank virtually without a trace. I have not seen a single significant reply to that short book from any of the scholars whose line of work Ellis’s book utterly demolishes. In a sense, these unhappy developments are not surprising. The problem with trying to refute a body of work in the humanities, particularly in literature,
. Andrew Ross, comment in a lecture at the New York Academy of Sciences, February 7, 1996. Cited by Alan Sokal,
, last consulted 12/2/2007. . The only review I have been able to find that is even arguably of any substance adroitly skips over what I believe to be Ellis’s central claim: that deconstruction proceeds from a theory of language, and that Derrida and his followers have deeply, grievously, perhaps even deliberately misread their theoretical base, Saussure’s Course in General Linguistics. See Norris (1990).
Donald C. Freeman
is that the concepts are very slippery, and literary study has little if any tradition of building upon a previous generation’s work. As a result, there is almost no philosophy of knowledge about literary study. Science, however, has a strong tradition of building upon existing foundations, and an entire discipline, the philosophy of science, devoted to what should count as a scientific fact, scientific argumentation, scientific method, and so on. So two scientists who believed scholars of “science studies” were ludicrously ignorant of basic science attacked “science studies,” in a book called Higher Superstition (Gross & Levitt 1994). This critique drew some attention in the academic world, but virtually none outside it. About a decade ago, however, this situation changed dramatically. A physicist named Alan Sokal wrote an article, “Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity” (Sokal 1996a), that appeared to be a practicing physicist’s conversion to the cause of “science studies.” Among the bolder claims in Sokal’s essay is his assertion that the famous constant of Euclidean plane geometry, π, is a historically contingent variable. Sokal submitted this farrago to the editors of Social Text, who published it without demur – and, significantly, without seeking the advice of a real scientist, as opposed to a practitioner of “science studies.” Sokal’s essay appeared in Social Text as a regular article, part of a special issue on the so-called “science wars.” Alas, simultaneously, and without telling the editors of Social Text, Sokal published in the gadfly (and now defunct, sadly) journal Lingua Franca an essay (Sokal 1996b) exposing his “Transgressing the Boundaries” article as a complete hoax. He adopted this strategy, Sokal wrote (1996b: 62), to show that “a leading North American journal of cultural studies … [would] publish an article liberally salted with nonsense if (a) it sounded good and (b) it flattered the editors’ ideological preconceptions.” His Social Text article was, Sokal wrote, a parody of “science studies” that was, in his own words (Sokal 1999: 268–269), “a mélange of truths, half-truths, quarter-truths, falsehoods, non-sequiturs, and syntactically correct sentences that have no meaning whatsoever.” This tempest burst forth from the academic teapot into a story carried on the front page of the New York Times one Saturday morning. A torrent of defenses, counterattacks, and counter-defenses ensued, culminating in a brilliant essay in the New York Review of Books (Weinberg 1996) eviscerating “science studies,” written by Steven Weinberg, a physicist and winner of the Nobel Prize. Dismayingly, Social Text’s editorial standards in the Sokal Affair attracted many staunch and prominent defenders. In an essay on the op-ed page of the New York Times, the ubiquitous Fish (1996) assailed Sokal’s demonstration as an ethical lapse and sought to defend “science studies” by comparing the rules of physics to the rules of baseball. More sweeping defenses could be heard in academic corridor chat. A common theme was that Sokal’s action had “damaged interdisciplinary research.”
Chapter 3. Notes towards a New Philology
What the Sokal affair damaged, of course, is bad interdisciplinary research. Equally disturbing was the rhetorical success of Social Text’s defenders in damage control, limiting the scope of Sokal’s critique to “science studies” in an effort to obscure the larger issue of slipshod interdisciplinary humanistic research in general. The thread of “interdisciplinarity” connects “science studies” to virtually all of the remaining “theoretical” enterprises: the linguistics of deconstruction; the law of “legal studies,” and the anthropology and sociology of “cultural studies.” The Sokal Affair demonstrated with appalling clarity that the “theory” undertaking has been intellectually flawed from the start. These events constitute a cautionary narrative. Its moral is: bad things happen when a bunch of mutually validating smart people think that they are above the rules. These rules include at least the following obligations: to make our work and its premises clear to our non-specialist peers and to the public; to answer serious critiques of our work in serious and non-dismissive ways; to be willing in principle to modify or abandon positions when we cannot answer these critiques; to protect the right of our intellectual adversaries to teach and publish their views, especially when they oppose our own; to pronounce as experts only where we possess expertise; and, most importantly, to police our own disciplines by calling to account the half-baked and the meretricious. Notwithstanding the jeremiadic tone of the foregoing, however, there is room for guarded optimism. Some of “theory’s” best and most ardent practitioners have begun to question assumptions in which their careers are heavily invested. Frank Lentricchia, the quondam “Dirty Harry of contemporary literary theory,” has, without explicitly saying so, retracted most of his own “theoretical” work. Lentricchia writes (1996: 65): If the authority of a contemporary literary critic lies in his theory of x, then wherein lies the authority of the theory itself? In disciplines in which he has little experience and less training. The typical literary critic who wields a theory is not himself a sociologist, historian, or economist, as well as a student of literature. A scandal of professional impersonation? No, because the impersonators speak only into the mirror of other impersonators and rarely to those in a position to test their theories for fraudulence. An advanced literature department is the place where you can write a dissertation on Wittgenstein and never have to face an examiner from the philosophy department. An advanced literature department is the place where you may speak endlessly about gender and never have to face the scrutiny of a biologist, because gender is just a social construction, and nature doesn’t exist.
. “Dirty Harry” was the sobriquet of a film character, a highly unorthodox San Francisco police lieutenant, played by the American movie star Clint Eastwood in several films in the 1970s and 1980s. The films immortalized the phrase “Make my day.”
Donald C. Freeman
Exactly. Even Fish, a brilliant critic who nevertheless must bear a large share of responsibility for the present quandary of literary scholarship, once confessed (Fish 1995: 110) that he “like[s] savouring the physical ‘taste’ of [literary] language at the same time that [he] work[s] to lay bare its physics.” The convergence of these developments and the Sokal Affair suggests that the time is ripe for a regeneration of literary scholarship using analytical methods, some of which are new and many of which have been around for a while but virtually ignored by mainstream literary research. This body of work makes it possible to develop real theory: accounts of literary works, oeuvres, genres, periods, etc., that are in principle predictive, explanatory, and falsifiable. Real theory is crucial to what I want to call the New Philology; at the end of this essay I will elaborate what I see as its crucial aspects. I focus here on one such methodology, cognitive metaphor (sometimes called “cognitive poetics,” [see M.H. Freeman 2005], a locution I find a trifle grand). Cognitivists argue that metaphor is a primary mode of thinking that is prior to and not restricted to language. On this argument, metaphor is constructed as schematized, embodied, and enculturated experience – spatial stories, on a recent account (originally in Turner [1996: 13–15]; for a more recent view, see Fauconnier & Turner [2002: 92–111]) – that is projected into abstractions. Consider, for example, the many ways in which we think about the abstraction we call “life.” One important way is to think of life as a journey along a path. Cognitivists say that we project or map a skeletalized mental representation, a schema, of the elements and structure of our physical experience of journeys into the abstraction “life.” Those elements are a beginning, an end, a route or path for the journey, something that moves (called a “trajector”), and a vector of progress. The structure would include the fact that the path has margins and that the normal progress along the path is forward from beginning to end (an account of the life is a journey metaphor is to be found in Lakoff & Turner [1989: 80–83]). Evidence for this analysis is found in idioms like “he’s reached the end of the road” (meaning “he has died”), notions of our lives “getting sidetracked,” that we’re “getting on in years,” the idea of “career paths,” “tenure tracks,” and so on. But there is literary evidence, too, in abundance: the first line of Dante’s Divine Comedy (“Midway in the journey of our life”), Robert Frost’s “The Road Not Taken,” Emily Dickinson’s “Because I could not stop for Death,” and so on. The claim is that one of the terms in which we think of life is that of a journey. Or consider the many different abstractions that we conceptualize in terms of our schematized, embodied experience of containers. Containers consist of a bounded periphery with an inside and an outside. We often think of moods and . As with interdisciplinary research in the humanities, the problem with “theory” has not been the idea of theory, but bad practices in the theoretical enterprise – “bad” in the senses I will suggest below.
Chapter 3. Notes towards a New Philology
states of affairs as containers: we are “in a bad mood” or “get into [and out of] trouble.” There is no a priori reason why we should think of moods or states of affairs in terms of containers. Yet we find ourselves “struggling to get out of bad relationships,” we go “into” and “out of ” debt, etc. Debts and bad relationships inhibit our ability to act freely. In cognitive terms, they constrain our freedom of movement; we find it difficult to get from the inside to the outside of their containing periphery. Cognitivists claim that our understanding of these abstractions is not arbitrary, but consistent with the independently motivated idea of metaphorical projection from our schematized, embodied experience of restriction and containment into our frustrated desire to escape what limits our freedom of movement. We map this physical experience into an otherwise unstructured, abstract idea of the emotional state called a mood (a full account of the container schema is to be found in Johnson [1987: 21–23]). I want to demonstrate how these ideas become a program for “real” literary theory by reexamining part of an analysis I wrote some years ago on one of the most analyzed speeches in literature, the “Tomorrow and tomorrow and tomorrow” speech in Macbeth: To-morrow, and to-morrow, and to-morrow Creeps in this petty pace from day to day To the last syllable of recorded time, And all our yesterdays have lighted fools The way to dusty death. Out, out, brief candle! Life’s but a walking shadow, a poor player That struts and frets his hour upon the stage And then is heard no more. It is a tale Told by an idiot, full of sound and fury, Signifying nothing.
V.v.19–28.
By this point in Macbeth the metaphorical projection life is a journey, part of the path schema, is well established in the play’s dramatic language. In Act I, when Duncan anoints Malcolm as his successor, Macbeth remarks: The Prince of Cumberland–that is a step On which I must fall down or else o’erleap, For in my way it lies.
I.iv.48–50.
. Freeman (1995). See this essay for a fuller account of the analyses I sketch out here of earlier speeches in the play. . All citations are from Barnet (1987).
Donald C. Freeman
In the depths of his despair in Act IV, the protagonist observes: My way of life Is fallen into the sear, the yellow leaf, And that which should accompany old age, As honor, love, obedience, troops of friends, I must not look to have….
IV.iii.22–26.
Here Shakespeare – typically, I suggest – blends four metaphorical projections, bad is down (the opposite of good is up, part of the verticality schema), life is a year, people are plants, and life is a journey. This last metaphorical projection Shakespeare – again, typically – manages to evoke a second time with one word, “troops,” where old age is seen as a kind of triumphal parade with troops of friends passing in review – along a path. The container schema is likewise well established by Act V. Lady Macbeth has remarked of her husband in Act I that he is “too full o’the milk of human kindness/To catch the nearest way.” (I.v.14–16). She understands Macbeth’s body as a container full of the wrong liquid. She would also change the liquid that fills the container of her own body, and seal it: Come you spirits That tend on mortal thoughts, unsex me here, And fill me from the crown to the toe top-full Of direst cruelty. Make thick my blood; Stop up th’access and passage to remorse, That no compunctious visitings of nature Shake my fell purpose nor keep peace between Th’effect and it.
I.v.38–45.
Both the path and container schemas are crucial in the “To-morrow” speech. There, the path along which time once traveled so freely for Macbeth has become contained within a “petty pace.” Shakespeare projects the schematized, embodied experience of containers onto the abstraction of depressed frustration. Macbeth is inside the container of his crimes and their consequences, vainly seeking to escape to the outside. These consequences are in the future, and it is that future that constrains Macbeth to the “petty pace” of those “tomorrows.” Time and Macbeth march in measured steps along each point, each day, of the “way to dusty death,” which is as inevitable an end to that path and that journey as the pen of a civil servant recording a legal document, left to right, syllable by syllable, until the end of time. Two vague measures of time now become pluralized and reified in another iteration of the life is a journey metaphor. As they “light fools the way to dusty death,” our yesterdays (the source point in the path schema) illuminate a path forward that is now constrained by the clearly visible terminal point of “dusty
Chapter 3. Notes towards a New Philology
death.” As the “fools” of humankind inevitably march toward that unmoving terminus, their – and Macbeth’s – path becomes a shrinking container, as the metaphors life is light and life is a journey become fused. Understood as cognitive metaphor enables us to understand it, this passage depicts Macbeth’s state of mind and situation in terms of a projection from a complex of very simple, ordinary, embodied experiences of journey, path, container, and light. Shakespeare lays out a precise horizontal spatial tableau of human mortality (the “fools”), life itself (the lamp held by “all our yesterdays,” for life is light), the trajectory of our lives (“the way”), and our lives’ containment by the inevitable terminal point of “dusty death.” This horizontal spatial tableau now – in four words, “Out, out, brief candle!” – is rotated 90 degrees to the vertical, with those entities and relationships intact. The path now has its source point not in the illuminating lamp of “all our yesterdays” but in the flame of the candle. That flame of life – for again, life is light – is the trajector, the moving entity, like the “fools” of l. 22. The flame now moves vertically, down the brief path from its present location in the candle’s wick to its extinction at the unmoving terminal point of the candle’s base, just as the “fools” inevitably move horizontally toward the unmoving terminal point of “dusty death.” Life is still a journey, but that journey now is down the candle – and bad is down. The candle, like the cone of light thrown by the illuminating lamp of “all our yesterdays,” is a bounded object. When the candle goes out, darkness will fall, and if life is light then death is darkness. But that is not the end of this story. As candles burn down, they cast flickering shadows. The steps of that shadow are, like the “to-morrows” of the speech’s beginning, constrained to the very short distance that an actor can “strut” (itself a short and constrained step) upon a stage, which is a constrained locus, and for a very short time, much less than the one-day minimum implied by “to-morrow” and “our yesterdays.” Finally, Shakespeare’s Macbeth invokes the common metaphor life is a story, describing life as a “tale,” one of the simplest prose literary forms, prototypically a straightforward narrative line without flashbacks or subplots – a narrative form that arguably partakes of the path schema. But the tale of Macbeth’s life is a “tale told by an idiot,” and tales told by idiots lack a coherent time scheme – they are journeys without coherent beginnings and ends. Macbeth’s mature career is, finally, a narrative whose path is not a straight line but one that is meaninglessly contorted and convoluted, a path of life in which what should have come at its end (“honor, love, obedience, troops of friends” [IV.iii.25]) came at its beginning, at a time when we think the natural movement is upward from where we are. Taken together, the container and path schemas interact in this speech to create a four-dimensional idealized cognitive model of Macbeth’s downfall: the path of his career becomes a container that constrains him in height (he can only “creep” and “strut,” he can no longer “o’ erleap,” as he does earlier in the play),
Donald C. Freeman
that constrains him in width (the “syllables” of time are recorded – and limited – horizontally; the actor-trajector in life’s drama can “strut” over no wider an area than a theatrical stage), that constrains him in depth (the “dusty death” of his end is now clearly lit and visible), and that constrains him in time (Macbeth’s “yesterdays” impel him toward a now enumerable and finite set of “tomorrows”). There are reasons why this speech is one of the most quoted and analyzed passages in the literature of the world – and I believe that a cognitive-metaphoric analysis provides a perspicuous account of those reasons. This kind of intense microanalysis is by no means all that cognitive metaphor can tell us about this play. Macbeth can be regarded as a kind of Gesamtkunstwerk, a total work of art (a term famously used of Richard Wagner’s operas), in which each part is structurally related to every other part. All of the great variety of elements that make up the prototypical Wagnerian opera fit together: Teutonic myth, symphonic (rather than operatic) orchestral accompaniment, massive choral singing, elaborate sets and costumes of a richness and complexity previously associated only with stage drama. Wagner scored for orchestras and choruses that were much larger than the customary size employed in operatic productions. He even designed an instrument called the Wagner tuba to get just the right brass sound, the distinctive voice that we hear, for example, in the Ring cycle of operas. Wagner even wrote his own libretti and scenarios, and composed the music that fit them. He designed and built a special opera house, the Festspielhaus in Bayreuth, that would give maximum effect to his totalizing operatic vision. In Wagner’s greatest operas, just about everything fits. In Macbeth, too, just about everything fits. It is the shortest of Shakespeare’s major plays; it is usually regarded as the densest. And I believe that cognitive metaphor as one basis for real literary theory can demonstrate the concentrated nature of Macbeth more persuasively than any other theory I know. For once we accept that metaphor is not a matter of language but a matter of thought prior to language (for an elaboration of this point see, among others, Turner [1996]) – we can see that the interaction of the path and container schemas I have described captures not only the language of the passages I have analyzed, but many other elements of the play. For example, by the time the play gets to the “Tomorrow” speech, we have already seen Macbeth’s career conventionally metaphorized as a journey along a path. But the beginning of Macbeth’s final downfall also invokes the path schema. Birnam Wood travels a path toward its terminal point of Dunsinane. Lady Macbeth sleepwalks – like that “tale told by an idiot,” the path of her journey is deranged: it has no coherent beginning or end. Shakespeare portrays Duncan’s deathbed as being in a room contained in a castle contained within a wall, strongly foregrounding each of these elements. Macbeth himself remarks at dusk on the
Chapter 3. Notes towards a New Philology
night he murders Duncan that “light thickens,” as though daylight had been boiled down in the container of the witches’ cauldron whose contents are simmered, as they put it, to a “gruel thick and slab.” And Macduff as the embodiment of retribution brings the container and path schemas full circle. He finally forces Macbeth literally to reverse direction on the path of his life (“Turn, hellhound, turn”). In his birth by Caesarean section, Macduff leaves the container of his mother’s womb “untimely ripp’d” from the conventional childbearing path (metaphorized in English as “the birth canal”) when he begins the journey of his life. Macbeth is a Gesamtkunstwerk. While this analysis doubtless could be improved, I know of no close-grained study of this speech’s language that connects as much of that language as does this analysis to the play’s larger issues and other dramaturgical elements: its plot, its characters, the structure of particular events. The explanatory power of analyses like this one arises from the theory of language upon which they are based. Cognitive metaphor is an important part of what I call the New Philology. The New Philology would assert for the present work that literary criticism of dramatic poetry begins with a rigorous analysis of its core metaphorical projections, along the lines of the foregoing claims for path and container metaphors as crucial to a reading of Macbeth, part of which I have articulated here; in Othello, the knowing is seeing metaphorical projection and its progeny (see Freeman 2004); in King Lear, balance metaphors (see Freeman 1993); and, in Antony and Cleopatra, metaphors of container, links, and path (see Freeman 1999). This methodology most highly values those analyses that give the deepest and broadest account of those projections and their operation in both the play’s language and in its plot, characterization, stage business, stage properties, etc. I call this work the New Philology to differentiate it from the traditional Germanic philology of my graduate training, a body of knowledge that was already on the way out of graduate English curricula in the U.S., but that I managed to experience before its demise. That era, the late 1950s, marked the end of a time in which advanced students in literature – English literature, at any rate – had to study English language as well as English literature, and were obliged as a part of their professional training to have first-hand acquaintance with English and Germanic philology and the then-contemporary analytical tools, knowledge, and theoretical insights available for their study. Despite the vast changes in the fields normally comprised within traditional philology, justifications for its inclusion in doctoral training have not weakened; on the contrary, the Chomskian and postChomskian theoretical revolutions in linguistics and the huge expansion of our knowledge about the human faculty of language greatly strengthen the case for requiring that advanced literature students study English language in its contemporary theoretical paradigms – for its substance, to be sure, but just as important,
Donald C. Freeman
for the intellectual style of modern linguistics. Linguistics has traditions of frank and spirited debate; of fundamental discussion of what constitutes a theory and how a theory can be supported or falsified; of great respect for facts and for evidence; and, despite the strongly left-wing political orientation of many prominent linguists, a traditional reluctance to equate particular scholarly approaches with personal politics. Contemporary doctoral students in literature should have a more than nodding acquaintance with contemporary theories of language: syntax, phonology, semantics, pragmatics, discourse. The decline to virtually zero of philology – the broad range of linguistic fact and theory that was once crucial in the training of literature scholars– has occurred pari passu with the ascendancy of “theory.” At least in the English-speaking world, the professional training and now the paradigms of research and publication in literary study have fallen prey to an intellectual Gresham’s Law: bad ideas have largely driven out good. Ideas that can be parodied so successfully that the parody, Sokal’s Hoax, fools self-proclaimed experts – are bad ideas. Ideas whose proponents will not answer serious critiques – are bad ideas. Ideas expressed in deliberately and defiantly, even proudly, obscurantist language – are bad ideas. Ideas that have been shown to be founded on fundamental errors in the disciplines of which their proponents profess knowledge – are bad ideas. But it does not suffice to bemoan these developments, even though Sokal’s Hoax and its aftermath demonstrated that growing academic and public concern about the decline of the humanities is well founded. What is required is a constructive and serious alternative program. I believe that the New Philology is such a program and that it can help redirect literary scholarship toward insights that are rigorous, falsifiable, and humanized. Sokal’s Hoax was a wakeup call – but too many literature scholars have yet to heed it. Thanks to thirty-plus years of “theory,” the academic study of literature has suffered significant damage, the full consequences of which are only beginning to be realized.10 I believe that the New Philology offers a promising basis upon which to reconstruct the study of literature as an academic discipline: literary analysis and criticism whose merits do not depend on its author’s politics; literary analysis and criticism that is open, explicit, and arguable; literary analysis and criticism that is, in the best sense, real literary theory.
. See, for example, the exchange between Donald Morton and a number of interlocutors in pmla 111 (1996): 133–134, 470–472. . For a similar account from a different perspective, see Cain (1996). His title alone, “A Literary Approach to Literature: Why English Departments Should Focus on Close Reading, Not Cultural Studies,” is astonishing. Imagine an essay in Science entitled “A Biological Approach to Biology.”
Chapter 3. Notes towards a New Philology
References Barnet, S. (ed.). 1987. Macbeth. New York NY: Signet New American Library. Cain, W.E. 1996. A literary approach to literature: Why English departments should focus on close reading, not cultural studies. Chronicle of Higher Education, December 13, 1996: B-4–5. Ellis, J. 1989. Against Deconstruction. Princeton NJ: Princeton University Press. Fauconnier, G. & Turner, M. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York NY: Basic Books. Fish, S. 1989. Doing What Comes Naturally. Durham NC: Duke University Press. Fish, S. 1994. There’s No Such Thing as Free Speech, and It’s a Good Thing, Too. Oxford: OUP. Fish, S. 1995. Why literary criticism is like virtue. In Professional Correctness: Literary Studies and Political Change, 93–114. Oxford: The Clarendon Press. Fish, S. 1996. Professor Sokal’s bad joke. New York Times (May 21, 1996), 23. Freeman, D.C. 1993. ‘According to my bond’: King Lear and re-cognition. Language and Literature 2(1): 1–18. Freeman, D.C. 1995. ‘Catch[ing] the nearest way’: Macbeth and cognitive metaphor. Journal of Pragmatics 24: 689–708. Freeman, D.C. 1999. ‘The rack dislimns’: Schema and metaphorical pattern in Antony and Cleopatra. Poetics Today 20(3): 443–60. Freeman, D.C. 2004. Othello and the ‘ocular proof ’. In The Shakespearean International Yearbook, G. Bradshaw, T. Bishop & M. Turner (eds), 56–71. Aldershot, UK: Ashgate Press. Freeman, M.H. 2005. Poetry as power: The dynamics of cognitive poetics as a scientific and literary paradigm. In Cognition and Literary Interpretation in Practice, H. Veivo, B. Pettersson & M. Polvinen (eds), 31–57. Helsinki: Helsinki University Press. Gross, P.R. & Levitt, N. 1994. Higher Superstition: The Academic Left and Its Quarrels with Science. Baltimore: The Johns Hopkins University Press. Johnson, M. 1987. The Body in the Mind. Chicago IL: University of Chicago Press. Lakoff, G. & Turner, M. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago IL: University of Chicago Press. Lentricchia, F. 1996. Last will and testament of an ex-literary critic. Lingua Franca 6(6): 59–67. Norris, C. 1990. Limited think: How not to read Derrida. Diacritics 20(1): 17–36. Sokal, A. 1996a. Transgressing the boundaries: Toward a transformative hermeneutics of quantum gravity. Social Text 46/47: 217–252. Sokal, A. 1996b. A physicist experiments with cultural studies. Lingua Franca 6(1): 62–64. Sokal, A. 1999. Transgressing the boundaries: An afterword. In Sokal & Bricmont 1999: 268–280. Sokal, A. & Bricmont, J. 1999. Fashionable Nonsense: Postmodern Intellectuals’ Abuse of Science. New York NY: Picador. Turner, M. 1996. The Literary Mind. Oxford: OUP. Weinberg, S. 1996. Sokal’s hoax. New York Review of Books (August 8, 1996), 11–16.
chapter 4
A theory of expressive reading Don Kuiken Whether justified by the “affective fallacy” or the “death of the subject,” challenges to expressivist theories of literary reading have been persuasive. What seems lacking is theory that respects the fragility of felt meanings and the vitality found in their uncovering. Addressing this lack requires articulation of how feeling expression unfolds over time, has the character of disclosure, and simultaneously brings feelings and their intentional objects to presence. Without detracting from expressive disclosure, it is also critical to acknowledge the limits of expressibility – and that expressive reading gestures toward the mood of that which cannot be brought to presence. Keywords: expression, disclosure, phenomenology, foregrounding, identification, feeling
1. Introduction In the current intellectual climate, it seems obligatory to speak in hushed tones about the possibility that literary reading entails significant moments of feeling expression. The history of this obligation is not simple – partly because it is theoretically over-determined and partly because its origins transcend theoretical perspective. For an earlier New Critical generation, declaration of the “affective fallacy” (Wimsatt 1954) was an attempt to distinguish the a-meaningful, nonreferential effects of the text from the meaningful, referential tensions or resistances within it (Brooks 1958). This resistance to epistemic subjectivity (Ogden & Richards 1923) invoked a subject-object split that, in practice, was maintained by shunning the feelings that emerged during even carefully considered interpretive efforts (Freund 1987). For a more recent generation, the episteme guiding Critical Theory also undermined the integrity of feeling expression – although in a quite different way. Assertions about the “death of the subject,” originally grounded in the Derridean critique of Husserl’s account of the living present (Derrida 1973),
Don Kuiken
left readers dangling somewhere between a radically deconstructed (and affectively neutered) subjectivity (Jameson 1991) and self-deceptive reversion to a nostalgic theory of feeling expression (Terada 2001). But, does feeling expression through literary reading inherently entail such interpretive and self-referential risks? Or, might feeling expression, when properly understood, provide access – even though vulnerable access – to the experiential potential of literary reading? Feeling expression has been construed in different ways, perhaps because, in actual fact, it occurs in several contrasting forms. One form is the unintentional objectification of an internal affective state (e.g., a blush), perhaps one that has been suppressed or defensively obscured (e.g., a slip-of-thetongue; Freud 1910/1957). Such symptoms of ongoing feelings motivate beliefs about an underlying affective state, both in the person whose feeling it is and in others who notice those symptoms. A second form of feeling expression occurs when objectifications of an internal affective state (e.g., gestures, utterances) deliberately convey something about that state for others to understand, respond to, or identify with (Tolstoy 1898/1995). Such communication of ongoing feelings assumes that the person conveying the feeling already understands their distinctive characteristics. A third form of feeling expression occurs when communication of ongoing feelings changes their characteristics (e.g., abreaction; Breuer & Freud 1893/1955; Kennedy-Moore & Watson 1999). Such modulation of ongoing feelings presumes not only prior understanding of their characteristics, but also tacit understanding of the consequences of their communication. However, symptomatic, communicative, and modulating forms of feeling expression are seldom at issue in theories of literary reading. Instead, following theories of poetic imagination (e.g., Dilthey 1887/1985), expression theories usually present the reader as actively moving toward the articulation of initially vague feelings. For example, in analytic aesthetics, Collingwood (1938: 109–110) argued that feeling expression begins with “oppressed” awareness of a vaguely sensed feeling and, through articulation, proceeds toward more complete awareness of that feeling’s characteristics. In continental aesthetics, Lyotard (1977: 13) spoke of “differend” as an “unstable state” in which the reader suffers from not being able to put something into words, a state that calls for new idioms to “express what is disclosed by the feeling.” Despite their contrasting modern and post-modern objectives, both Collingwood and Lyotard refer to a form of feeling expression that provides an unfolding movement toward uncovering, or disclosure, regardless of whether what is disclosed represents something presentable (Collingwood) or gestures toward something that is ultimately unpresentable (Lyotard). However, what is expressively disclosed is not limited to the interiority of feeling; rather, feeling expression as disclosure can – and often does – occur
Chapter 4. A theory of expressive reading
precisely where expression and representation coincide. Thus, rather than an extraneous effect of the aesthetic object, feeling expression may be intrinsic to aesthetic understanding. Within analytic aesthetics, conceiving expression and representation as coincident is uncommon (Scruton & Munro 2003). Even so, Goodman (1976), a noteworthy exception, proposed that expression, including feeling expression, can be understood as metaphoric exemplification, a complex form of denotative reference. For him, a dance that metaphorically exemplifies light heartedness expresses light heartedness, affording epistemic access to those feelings independently of whether they are engendered in the respondent. Within continental aesthetics (Bowie 1997), the occurrence of disclosure is often construed as an event within which feeling expression and disclosing representation go hand in hand. Heidegger (1927/1996), for example, argued that the disclosure of beings unfolds within an attunement to the world (Stimmung) that has the character of what is commonly called a mood (Befindlichkeit). A work of art creates a clearing within which the pre-understanding implicit in such a mood brings something to presence, something that lingers or endures. In sum, within both analytic and continental aesthetics, there are conceptions of feeling expression that present it as unfolding over time, having the character of disclosure, and simultaneously bringing feelings and their intentional objects to presence. Further articulation of these conceptions, with particular attention to literary reading, can be pursued by drawing on discussions within the continental tradition, with accents provided by Eugene Gendlin (1962/1997; 1997), a philosopher-psychologist who has written especially vividly about the vicissitudes of feeling expression.
2. The character of feeling In a theory of feeling expression as disclosure, it is important to be as explicit as possible about what is meant by feeling. It will not suffice to construe feeling as merely the reflective perception of bodily states, even if those bodily states involve intricately patterned kinaesthesia or proprioception. Feeling may include these afferent sensations from the physical body but it is not limited to them. Instead, feeling also is reflectively found in the felt sense of the lived body’s projected actions (Merleau-Ponty 1942/1962). Gendlin (1992; 1997) emphasizes that these projected actions involve interaction with an environment and that the complexity of interactional living is sensed in the gestures, actions, and utterances that we potentially perform. By extension, feelings also may be sensed in the physical body and in the lived body during the response-organizing activities of basic emotions,
Don Kuiken
such as fear, anger, and sadness, as well as curiosity (Ellis 2005; Panksepp 1998). However, the felt aspects of emotion are only one aspect of the more inclusive conception of feeling identified here. The preceding sources circumscribe what might be called personal feeling, but there is ample reason to resist the impression that feeling is simply internal. What is felt does not stop at the boundaries of the body, as indicated by our capacity to sense the space behind our backs (Merleau-Ponty 1945/1962) or, in dance, to shape bodily a space that is thereby given a mood. Also, feeling can be reflectively found when objects beyond the body boundary are taken up as tools in the service of skilful knowing. For example, the “feel” of a tool in one’s hand suggests that it has been incorporated into in an overall sense of one’s lived body – an incorporation that Polanyi (1964) called “indwelling.” Finally, feelings can reflectively be found in what Merleau-Ponty (1945/1962) refers to as the reversibility of subject and object, i.e., the enfolding of the sensor and the sensible within a single embodied grasp of the object. As described in one of his examples: As I contemplate the blue of the sky I am not set over against it as an acosmic subject; I do not possess it in thought, or spread out towards it some idea of blue such as might reveal the secret of it, I abandon myself to it and plunge into this mystery, it ‘thinks itself within me’, I am the sky itself as it is drawn together and unified, and as it begins to exist for itself; my consciousness is saturated with this limitless blue. (Merleau-Ponty 1945/1962: 214, original italics).
In other words, the inner “flesh” is experienced as the “flesh” of the world (Merleau-Ponty 1977), and recognizable feeling is extended to the objects of perception, establishing what might be called their felt presence (Kuiken 1998). The experience of any particular situation will be comprised of a blend of the preceding forms of reflectively found feeling. They jointly contribute to the sense of being-in-a-situation, i.e., to the “felt sense” of that situation (Gendlin 1962/1997; 1997). By implication, they also contribute to the literary reader’s felt sense of situations in the world of the text.
3. Openness to feeling expression as disclosure Conceived as disclosure, feeling expression during literary reading calls for openness to the progressive uncovering of objects within the world of the text. But the kind of openness required is not simply openness to, or resilient tolerance for, the aversive import of emotional feeling (e.g., sadness, anxiety). Without diminishing the importance of such resilience, a subtler and more inclusive form of openness is
Chapter 4. A theory of expressive reading
required, in particular, a form that is traceable to Kant’s (1790/1951) conception of disinterestedness. Disinterestedness may be understood as a form of restraint that mitigates against premature closure and, by implication, supports openness to the temporally unfolding course of feeling expression. Some modern versions of aesthetic attitude theory (cf. Fenner 1994), while also rooted in Kant’s conception of disinterestedness, fail to retain distinctions that are important for a theory of feeling expression as disclosure – although these limitations are instructive. According to aesthetic attitude theory, when a reader suspends practical objectives (e.g., puts aside reading for social gain or for educational purposes), she is able to become more fully absorbed in the world of the text (Stolnitz 1960). One explanation (Dickie 1964) is that the suspension of practical objectives supports unfettered attention to aspects of the text that ground comprehensive and valid interpretation. But, framing the problem in this manner implies that the objects of aesthetic contemplation are preset and determinate and that the primary question is whether the reader attends to them. This construal misleadingly emphasizes obstacles to determinate judgment, rather than impediments to the disclosing dynamics of indeterminate judgment. Kant’s (1790/1951) discussion of disinterestedness centered on the latter: for him, an interest entails a determinate concept of an object of desire. In that respect, interest is an impediment to the disclosing dynamics of indeterminate judgment, and, in contrast, disinterested contemplation may be understood as openness to those dynamics. Stated differently, disinterested contemplation is open responsiveness during reflection on the preconceptual, unsayable “more” that is implicit in the felt sense of a situation. It is the restraint required to avoid premature closure and, instead, respond to whatever comes from reflection on the preconceptual “more,” on the unsayable “excess,” on the vaguely felt “edge” of the felt sense of a situation (Gendlin 2004). To appreciate the sense in which this openness involves restraint, it is important to distinguish: (1) reflection on the felt sense of “more” that is followed by recovery and recognition that the “more” has been said before (e.g., as in the tip-of-the tongue phenomenon); (2) reflection on the felt sense of “more” following which some aspect of the “more” becomes sayable for the first time (i.e., fresh disclosure); and (3) reflection on the felt sense of “more” following which what has not been said remains unsayable (i.e., a persistent sense that the “more” cannot be said). Pretending that one can know in advance the outcome of reflection on a vaguely felt sense of “more” is metaphysical presumption – whether it occurs among ordinary readers or literary theorists. In contrast, disinterested openness during reflection on the felt “edge” of a felt sense is the waiting readiness for either the recovery and recognition of what has been said before or freshly disclosing expression of the “more” or the impossibility of saying what seems implicit in the “more.”
Don Kuiken
This formulation echoes Heidegger’s (1953/2000) insistence on the “letting be,” the “waiting” that shelters the withdrawal of being – and his resistance to the willed intelligibility that does it violence. Disinterested openness to the dynamics of expressive disclosure is a restrained waiting for a saying that comes from the felt sense. What comes, the saying that emerges from the felt sense, is not always resonant with that felt sense; it may provide a mute “not this,” a silent “not that,” but also, sometimes, a fresh, responsive, and resonant saying in which the “more” that is implicit in the felt sense is “carried forward” (Gendlin 2004). In that case, the expressed meaning retains something of what was already implied, although it now also has a new intricacy that implies still “more” (Gendlin 1991). Such carryingforward, when it occurs, begins with sustained reflection on a vaguely felt sense of a situation and moves toward a fresh, vital expression of what is implicit in it. Consistent with the preceding formulation, there is evidence (Kuiken, Busink, Miall & Cey 2003) that literary readers who have been given instructions supporting covert creation of a quiet and protected “space” for reflection are more likely to report (1) resonance of their own feelings with those expressed in the text, (2) an objective impression of the feelings expressed in the text, and (3) an experiential shift through which they carried forward feelings they had usually ignored. Moreover, readers who experience such shifts are those who imagine the text vividly, read for insight, and report that feelings in emotion-relevant areas of their bodies are easier to hold and capture than feelings in emotion-irrelevant areas (Kuiken, Busink, Dukewich & Gendlin 1996). Thus, the form of openness described here demonstrably influences readers’ readiness to entertain the dynamics of expressive disclosure. 4. Aesthetic feeling: Carrying forward a felt sense of … If disinterested openness allows broad access to the dynamics of expressive disclosure, “direct reference” (Gendlin 1962/1997) to the “more” that is implicit in a felt sense is the first moment in those dynamics. During literary reading, direct reference entails reflective “listening” (Nancy 2002/2007), a reflective turning toward the felt sense of “more” that momentarily – and periodically – emerges during the covert saying that is the text-as-it-is-being-read. In direct reference, the reader listens for a further saying that returns a modification of that felt sense of “more.” The anticipated modification, which comes from the reader’s own unfolding saying of the text-as-it-is-being-read, is resonant with, and not merely a . The original Kuiken et al. (1996) report describing development of this measure also provided evidence that this affective body awareness index predicted self-perceptual disclosure following impactful dreams.
Chapter 4. A theory of expressive reading
repetition of, the felt sense of the original saying. Thus, rather than listening for reactivation of the original felt sense, which Merleau-Ponty (2002) calls “spoken speech,” readers listen for a resonance that comes with regenerative saying, what he calls “speaking speech.” During literary reading, listening for the resonance of “speaking speech” often begins within the context of aesthetic feeling, i.e., during reflection on the felt sense of “more” that makes figuratively enriched (foregrounded) passages seem lively, striking, or evocative (Miall & Kuiken 1994; Mukarovsky 1932/1964; Šklovskij, 1917/1965; Sopcák 2007; Van Peer 1986). More than appreciation of the formal aspects of a text, aesthetic feeling, then, is a felt sense that initiates listening for the resonance that comes with covert regenerative saying. That covert saying will, as indicated earlier, sometimes be mute, lacking resonance. And, sometimes it will merely hold the sense of “more” in place, as though to provide a “handle” that enables returning and listening again for the resonance that might come (Gendlin 1996: 48). But, sometimes that covert saying will resonate and carry forward, however slightly, the intimate intricacy that seemed implicit within the “more” of the aesthetically felt sense. The intimacy of the newly said intricacy was already implicit in the felt sense of the figuratively expressed meanings. Evidence suggests that synaesthetic metaphors in literary texts move from distal (e.g., vision) to proximal (e.g., touch) sensory modalities (Ullman 1957; Shen 1997), that similes move from less to more concrete and accessible meanings (Shen 1995; 2007), and that zeugma join less with more concrete and accessible meanings (Shen 1998). These findings help to explain why the saying that comes from reflection on the felt sense of figuratively enriched passages discloses a more intimate felt sense. This intimacy is a distinctively embodied form of “refamiliarization” (Miall & Kuiken 1995; Fiahlo 2007) that displaces the initial “defamiliarizing” effect of the foregrounded passages (Miall & Kuiken 1995). Moreover, as Gendlin (2004) has emphasized, a “life-enhancing” vitality marks the carrying forward of implicitly felt meanings. In Kant’s (1790/1951) original – and related – account, the harmonious interplay between imagination and understanding, which creates a self-sustaining sequence of aesthetic ideas, is sensed as the “feeling of life” (“Lebensgefühl,” §1, 38), or, in slightly different terms, the “feeling of the furtherance of life” (“Gefühl der Beförderung des Lebens,” §23, 83). Gefühl der Beförderung (transportation, carriage) suggests the feeling of aesthetic ideas being “carried forward,” perhaps toward fuller, richer disclosure.
. Kant’s proposal was echoed later by Dilthey (1887/1985). Since Gendlin studied Dilthey early in his career, reference to the life-enhancing character of “carrying forward” may, in part, derive from that source.
Don Kuiken
The onset of these vital feelings of modification may also, as Gendlin (1996) suggests, have the character of “release,” as though there is a subtle “letting go” within the saying that constitutes expressive disclosure. An overly simple (and hypothetical) example may help to concretize the aspects of literary reading that have been described so far. In the following excerpt from Borges’ (1946/2000) essay, “A New Refutation of Time,” an abstract understanding of time is metaphorically given a more concrete and accessible sense, specifically, as a moving force: Time is a river that carries me along.
An altered sense of time is suggested by metaphorically enfolding it within a sense of motion with which it is not usually associated. In its first iteration (“Time is a river”), a simple metaphoric statement accentuates that sense of flow, drawing the reader away from the more conventional linear conception of time. But toward what understanding of time does this metaphor move? Is it toward time that flows? Is it toward time with depth? Disinterested listening (i.e., direct reference) to the felt sense of something “more” that has been accentuated by the initial metaphor involves anticipation of potential resonance either with the spontaneous explication that re-writes the written text (the covert supplemental saying of “time that flows,” “time with depth”) or with the explication that is given by the remainder of the written line (“that carries me along”; italics here suggest an emphasis specific to this reader’s covert saying of the text). If the explicative saying that resonates with the original felt sense of the metaphor is found “within” the text (“that carries me along”), the original felt sense will be carried forward and, with it, the fresh intricacy of supportive carriage or personal transport within a more intimate sense of time. The reader may find this modification of meaning enlivening, with an accompanying sense of release; there will also be a freshly felt sense of “more” that prompts listening for a somewhat different resonance than seemed “needed” before. The utility of this simple example depends partly on the way in which the meaning of its simple nominal metaphor is explicated; describing the disclosing potential of even this simple metaphor helps to articulate how a felt sense is freshly carried forward in response to figuratively enriched passages. Understood as mere comparison (where “Time is a river” is considered equivalent to “Time is like a river”), the generative potential of metaphor is obscured. Instead, a class inclusion theory that describes the generativity of novel metaphoric expressions is required (Glucksberg 2006). According to Glucksberg and Keysar’s (1990) class inclusion theory, in simple nominal metaphors the metaphoric vehicle refers through exemplification to an ad hoc class that includes both the metaphoric vehicle and topic. So, in our example, “river” exemplifies an ad hoc class that also includes conceptions of “time.” Such metaphoric reference facilitates consideration of
Chapter 4. A theory of expressive reading
the notion that “time” has the attributes of this ad hoc class (e.g., time flows), while inhibiting consideration of attributes that do not describe variations among members of the metaphoric topic (e.g., wetness is not a relevant attribute among conceptions of time). For many metaphors, the comparison and class inclusion theories yield the same predictions, but, for apt metaphors, comparison theories fail to account for the asymmetric metaphors that can be explained by class inclusion theories (e.g., “My surgeon is a butcher” is not equivalent to “My butcher is a surgeon”; Glucksberg 2006). In addition to the generativity implicit in the creative “naming” of an ad hoc class, apt metaphors are generative in that they are more likely to facilitate the consideration of emergent features, i.e., features of the metaphoric topic that become salient when reading the intact metaphoric expression but that are not salient when the metaphoric vehicle or metaphoric topic are separately considered (Becker 1997; Tourangeau & Rips 1991; Gineste, Indurkhya & Scart 2000). Thus, in our example, although neither “time” nor “river” have “carries me along” as a salient feature, “carries me along” is an emergent feature of time when considered in light of the metaphoric expression. The emergence of a freshly apt meaning, through the creation of new similarities between vehicle and topic, exemplifies the potential for expressive disclosure, for speaking speech, in literary reading (Utsumi 2002).
5. Theme variations: Transforming shifts in a felt sense of … The limited empirical evidence available (Miall & Kuiken 1995) indicates that the emergence of and response to aesthetic feeling is local and brief, providing (1) a fleeting accentuation of interest and evocative promise, (2) a brief period of deliberation about the sayings that come from that felt sense of “more,” and then (3) a reported shift in understanding that briefly guides further reading. However, these local dynamics enable the emergence of a second moment in the overall dynamics of expression: the emergence of affective themes. The aesthetically felt sense of a foregrounded passage guides the reader’s recognition of subsequent feeling-congruent passages (Miall & Kuiken 2002), facilitating the identification of recurrent affective themes. Such feeling-congruent passages may be best understood as having an affective family resemblance because their complexity is enhanced by the capacity of feeling to guide the detection . Among readers in our studies of short stories (Miall & Kuiken, 1995), heightened interest in response to foregrounding lasted only a few seconds, listening for the alternative understandings persisted for 15–20 seconds, and the felt shift in understanding most frequently peaked about 30 seconds after reading the foregrounded passage.
Don Kuiken
of feeling congruity across quite different narrative content. Rather than being limited to feeling congruity between one character and another or between one setting and another, theme variations often depend upon feeling congruity between characters and settings, characters and institutions, settings and events, etc. For example, the following reader commentary, taken from a study of readers’ response to Coleridge’s Rime of the Ancient Mariner (Sikora, Kuiken & Miall 1998), initially refers to the danger of enchantment in the wedding guest’s encounter with the Mariner: Passage # 1: He holds him with his glittering eye – // The wedding guest stood still, // And listens like a three years’ child: // The mariner hath his will. // The wedding guest sat on a stone: // He cannot choose but hear; (lines 13–18) Commentary: I like it because it appeals to me because of the … just knowing that stories do have that kind of power. There’s also an element of threat to it, like an enchantment, but there’s also an element of danger because he’s not there because he wants to be. He feels he has no control. I relate to this just because I have been known to get caught up in books or in stories. I love listening to stories, so I know that they do have just about that kind of power. You don’t want them to end; you have to hear what happens. Curiosity is so completely aroused.
This reader views stories as entailing the threat of being caught up in something from which she is powerless to escape. She identifies a similar threat in a subsequent passage – but now in response to a setting description: Passage #2: The ice was here, the ice was there, // The ice was all around: // It cracked and growled, and roared and howled, // Like noises in a swound!. (lines 59–62) Commentary: I just like the image, the feeling of nature being alive, like it’s ice but there’s more to it, that there’s a spiritual force. There’s again a threat … the feeling of being surrounded. It’s like a completely alien world that we’re not used to being in … of being surrounded by ice with no living thing around. It’s hauntingly beautiful and powerful but very frightening.
The locus of the threat in this commentary has changed from a “powerful” character to a “completely alien” setting but the affective theme remains similar: threatening enchantment. Themes that are constituted through such feeling congruent boundary crossings are “motifs of consciousness” (Befindichkeitsmotif), involving passions, moods, and attitudes, including moral attitudes (Wolpers 1995: 35). Rather than depending upon similarities in content (e.g., places, actions, life stages), affective themes develop from similarities in the freshly intimate and intricate aspects of each felt sense that has been separately carried forward while reading the foregrounded passages whose boundaries are “crossed.” Consequently, as demonstrated in the
Chapter 4. A theory of expressive reading
preceding commentary, even resonance between the aesthetically felt presence of animate beings (e.g., the Mariner as holding, threatening) and of figuratively enlivened inanimate beings (e.g., the ice as living, growling) may contribute to the emergence of a theme. The reader’s open listening to the covert reading of a literary text enables such resonance between the aesthetically felt sense of each of the separate passages that constitute the theme. Variations on a theme “cross” in a way that is different from the way vehicle and topic “cross” in metaphor (Gendlin 1995). Instead, a new dynamic contributes to the resonance that occurs between iterative variations on a theme. It may be useful to work again with a simple example. The previous excerpt from Borges’ (1946/2000) essay continues as follows: Time is a river that carries me along; It is a tiger that devours me.
Construing time as river and then tiger, this sequence of metaphors provides variations on a theme that enables time to accumulate additional meaning complexity. Through the form of listening described earlier, time as a river that “carries me along” is initially carried forward as a felt sense of supportive carriage or personal transport. Subsequently, time as a tiger that “devours me” is carried forward as a felt sense of terminal engulfment. But reading this second variation on the theme provides a crossing of these freshly sensed meanings that transforms them both. Three types of transformation are involved: • Metaphoric Feeling Transformation. An initial incongruity between time as supportive carriage and time as terminal engulfment evokes the possibility that time provides supportive carriage – but carriage toward terminal engulfment. This transformation has the character of composite metaphor (Lakoff & Turner 1989): metaphorically, the time that “carries me along” is the time that “devours me.” In this transformation, the time that devours is a metaphoric vehicle that carries foreword to disclosure the time that provides personal carriage toward terminal engulfment. • Saturated Feeling Transformation. Independently of metaphor, the succession of theme variations also provides a blending of compatible features into a composite so that the unfolding affective theme becomes saturated with layers of meaning. In this case, time retains its sense of personal carriage – even though now personal carriage, perhaps like that provided by a mightier river, is a source of powerful, uncontrollable movement. This composite, personal carriage and uncontrollable movement, is an integrative blend (Grady, Oakley & Coulson 1999), rather than a metaphoric fusion.
Don Kuiken
• Emotional Feeling Transformation. The succession of theme transformations also moves from emotional feelings of one kind to emotional feelings of another kind. The sense of time as calmly flowing carriage becomes time as terrifyingly devouring terminal engulfment. Rather than mere intensification, this shift is a qualitative change in the kind of emotion that is felt within an invariant construal of time as moving toward terminal engulfment. Although transformations analogous to these have been documented in studies of literary texts per se (cf. Zholkovsky 1984), unfortunately there has been no systematic empirical study of these transformations in accounts of actual reader’s experience. Metaphors of Personal Identification: Owning a Felt Sense of …
Theme variations seem inherently to move toward increased inclusiveness. That is, an affective theme becomes understood as pertaining first to one character, then to another, later to a setting and still later to an action sequence, and so on. Listening for the resonance across theme variations gradually discloses broadly person-relevant, and not only self-relevant, meanings. Thus, the generativity of theme variations will at times challenge the reader’s sense of self within a succession of expressive moments. However, one particular form of such self-implicating reading seems to facilitate feeling expression as disclosure. Consider the following commentary from a reader of The Wrong House, a short story by Katherine Mansfield, which we have employed in several of our empirical studies (e.g., Miall & Kuiken 1994). The study in which she participated involved a modified self-probed retrospection technique, and she marked and commented on Mansfield’s description of a story setting in which “the houses opposite looked as though they had been cut out with a pair of ugly steel scissors and pasted on to the grey paper sky”: I could actually see the street and the houses. So there was great imagery there … it reminded me of the street that I lived on when I was young. We lived in a small town in southern Alberta … and the houses looked like that; they looked like they had been cut out with ugly steel scissors.
As this example affirms, the personal memories evoked during reading often capture similarities between aspects of a personal memory and aspects of the world of the text. In this case, the comparison is explicit (“the houses looked like that”), which suggests that this reader’s expression can be understood as a simile (A is like B; my experience of the street in my home town is like the narrator’s experience of the street in the story). As implied by the simile, memory
Chapter 4. A theory of expressive reading
and story are symmetrical partners in a comparison (“A is like B” is equivalent in meaning to “B is like A”). In contrast, some readers describe a form of self-implication that resembles what Cohen (1999) has called metaphors of personal identification. In the following example, the reader was commenting on Mansfield’s description of a moment in which the protagonist, Mrs Bean, realizes that men on a funeral coach are disembarking and approaching her door: “ ‘ No!,’ she groaned. But yes, the blow fell, and for the moment it struck her down. She gasped, a great cold shiver went through her, and stayed in her hands and knees.” Our reader comments: it just makes you realize that … your own mortality is something that can make you unable to think clearly … [W]hile you think you still are alive and well and able to take care of yourself and help others, somebody else has decided that you can’t. And then [at times like this] you don’t think that it’s their problem, [but instead] that you somehow have been mistaken all this time and that it’s time for you to give in and end everything, whether you’re ready to or not … [A] passage like this makes you realize that some day, perhaps something like that will happen to you and scare the hell out of you because you know how close it could be for you.
This reader uses the pronoun “you” to speak inclusively but still personally (e.g., “it just makes you realize”). While spelling out what Mrs Bean is like in this scene (e.g., she was “unable to think clearly,” she is realizing that “it’s time … to give in”), this reader is also implicitly referring to herself as a person of the same kind; she is entertaining the possibility that she is the same kind of person as Mrs Bean. Although similarity is somehow at stake, this reader is not simply comparing Mrs Bean and herself. Although she identifies Mrs Bean and herself as members of the same inclusive class, they are not situated in the same way within that class. The pronoun “you” has a slight externalizing “otherness”; the description enlivens and extends her experience of the world of the text; and, in this way, she is brought to presence while the reader remains implicated but subsidiary. Mrs. Bean (but not the reader) is made to exemplify the class, and comparison between Mrs. Bean and the reader, in this context, is asymmetrical; that is, to say, as this reader seems to do, that, “I am Mrs Bean” is not equivalent to saying that “Mrs Bean is me.” Such asymmetry affirms that, rather than comparison through simile, this reader is engaged in a metaphor of personal identification (Cohen 1999). Consistent with the earlier discussion of metaphors in general, the reader’s identification of herself with Mrs Bean creates an ad hoc class exemplified by Mrs Bean but also including herself. This metaphoric self-reference implicitly uncovers in the reader those attributes of the ad hoc class exemplified by Mrs Bean
Don Kuiken
(e.g., those who are “unable to think clearly,” who realize that “it’s time … to give in,” and so on) within the constraints imposed by the reader’s own selfunderstanding (e.g., that she is not old like Mrs Bean). It also opens the way for expressive disclosure by prompting her to consider whether she possesses the previously unarticulated attributes of individuals in this class. Examination of the concluding passage in our reader’s reflections substantiates this possibility: she is made to “realize” that “something like that will happen to [her] and scare the hell out of [her] because [she will] know how close it could be.” One of the ten “critical difficulties” identified by I. A. Richards (1930) in his study of response to poetry was the relevance of personal associations, memories, and images. Although he was particularly concerned with the “intrusion” of these personal elements upon the “autonomy” of the text’s “meaning” (Richards 1930: 236), the issue of relevance remains important when considering the relevance of the preceding forms of self-implication. However, within metaphors of personal identification, Mrs. Bean remains the exemplar of an ad hoc category that is now tacitly understood to include the reader. The autonomy of the text is not at stake, but the joint understanding of Mrs. Bean and the reader is. What is expressively disclosed is not merely the interiority of the reader’s feelings; instead, expression as disclosure occurs at the confluence of expression and representation. Rather than infringing on the autonomy of the text, the feeling expression afforded by metaphors of personal identification seems intrinsic to aesthetic understanding of the world of the text. Returning to, and extending, the simple Borges example provides some indication of how metaphors of personal identification become enfolded within a succession of theme variations: Time is a river that carries me along, and I am the river; It is a tiger that devours me, and I am the tiger; It is a fire that consumes me, and I am the fire.
As these lines suggest, readers (with or without such explicit textual invitation) may become metaphorically identified with a succession of figures, not only with personified rivers, tigers, and fires, but also with characters, the narrator, and the implied author (Booth 1961). Iser (1978: 213), like Bakhtin (1981), has also emphasized the temporal course of self-implication in literary reading, suggesting that successive identifications with a character, narrator, or implied author are more plausibly understood as a dialogic sequence of identifications – a sequence that may not only contribute to the transformation of affective themes but also to disclosing modification of the reader’s sense of self.
Chapter 4. A theory of expressive reading
There is evidence that readers who experience reading as disclosure are those who both revisit affective themes during their reading and do so in ways thatinvolve metaphors of personal identification (Kuiken, Phillips, Gregus, Miall, Verbitsky & Tonkonogy 2004). That is, reported affective theme variations only lead to expressive disclosure when metaphors of personal identification persist as part of their structure. We have elsewhere described this style of reading as expressive enactment (Sikora, Kuiken & Miall 1998).
6. The limits of expressibility: Gesturing toward the unpresentable … If disinterested contemplation is open responsiveness during reflection on the unsayable “more” that is implicit in the felt sense of a situation, it is critical to remain attentive to the possibility that reflection on the felt sense of “more” will sometimes lead to a persistent sense that the “more” is unsayable. Although closure in attempts to articulate the limits of expressibility is, perhaps in principle, impossible, some aspects of those limits, as limits, are intelligible. Although time and space do not allow full development of this proposal, I suggest that the reader’s movement toward expressive disclosure through aesthetic feeling, theme variations, and metaphors of personal identification is, in fact, necessary to locate and gesture toward what cannot be disclosed. When unfolding expressive disclosure is disrupted, it is then that the figurative forms promising “more” lose the sense that there is “more” that can be said. The following lines from Wordsworth’s Resolution and Independence concretize this possibility: As a huge stone is sometimes seen to lie Couched on the bald top of an eminence; Wonder to all who do the same espy, By what means it could thither come, and whence; So that it seems a thing endued with sense: Like a sea-beast crawled forth, that on a shelf Of rock or sand reposeth, there to sun itself; Such seemed this Man, not all alive nor dead, Nor all asleep – in his extreme old age: His body was bent double, feet and head Coming together in life’s pilgrimage; As if some dire constraint of pain, or rage Of sickness felt by him in times long past, A more than human weight upon his frame had cast.
Wordsworth’s portrayal of the leech gatherer presents the old man near a pond, his body “bent double,” feet and head coming together under the weight of “life’s
Don Kuiken
pilgrimage,” comparable to a “huge stone” that is “couched on the bald top of an eminence.” The huge stone on an eminence, in turn, is compared to a “sea beast” crawling onto shore to “sun itself.” The images of the old man, the huge stone, and the sea beast coalesce and gesture toward an ontological category that transcends each separate image and that is not explicable but also not nonsensical, a category of that which is “not all alive or dead.” The enfolding of this three-fold figuratively modified image into an inexplicable ontological category transforms the promise of “more” that might be sensed within each of those images into an insufficient gesture toward an inexplicable category. If metaphor is the primary trope of disclosure, irony would seem to be the primary trope for gesturing toward that which inexplicably resists disclosure. “Not all alive or dead,” like the simpler oxymoron “living death,” might seem to encapsulate direct antonyms, i.e., two opposite poles on a certain dimension (hot/cold), two opposite values on a distinguishing feature (man/woman), or two irreconcilable attributes (square/circle). And yet, as Shen (2007) has shown, oxymora in literary texts most commonly occur in an indirect form (e.g., whistling silence), in which one of its terms (whistling) is a specific instance of a more general category (sound) that is the antonym of the other term (silence). In these cases, the specific instance of a more general category may be said to gesture toward the impossibility of the bringing the oxymoron’s antonymic terms to simultaneous presence. The oxymoron, then, seems the prototype of those ironic figurative forms that gesture toward the impossibility of saying more. Shen’s research (Shen & Balaban 2005) also suggests that indirect oxymora are often interpreted as though the specific instance of the general category (e.g., “watery” in “watery dryness”) metaphorically describes the antonymic phrase (e.g., “This dryness is water-that-floods-over-me”). This may explain why the interpretation of indirect oxymora often provides emergent associations (Gibbs and Kearney 1994) that seem to gesture toward the possibility of giving presence to an opposition the abstract poles of which cannot be given simultaneous presence. Perhaps, as Burke (1969) argued, the structure of irony contains aspects of the structure of metaphor, giving it the capacity to allude to, if not disclose, what is beyond saying. By implication, more fully theorizing the “more” that seems beyond the limits of expressibility will itself require a theory of expressive disclosure. Articulation of how the felt sense of “more” can sometimes be carried forward as disclosure enables more precise characterization of what happens when reflection on a vaguely felt sense of “more” affirms the impossibility of saying what seems implicit in the “more.” Articulation of the interplay between disclosure and its disruptions seems critical especially as empirical research ventures in the direction of sublime feeling (Kuiken, Miall & Aaftink 2006). The varying mood of sublime moments during
Chapter 4. A theory of expressive reading
literary reading, ranging from enthrallment to deep disquietude, only underlines that careful characterization of feeling expression and its vicissitudes will be necessary. Without such theory, it might be misleadingly argued either that the expressive subject dies beyond the limits of expressiblity or that the expressive subject lives only when feeling expression affords presence. Instead, we can look for the expressive subject wherever reflective reading initiates exploration of the equiprimordiality of presence and absence.
References Bakhtin, M.M. 1981. The Dialogic Imagination: Four Essays. M. Holquist (ed.), C. Emerson & M. Holquist (transl.) Austin TX: University of Texas Press. Becker, A. 1997. Emergent and common features influence metaphor interpretation. Metaphor and Symbol 12: 243–259. Booth, W.C. 1961. The Rhetoric of Fiction. Chicago IL: University of Chicago Press. Borges, J.L. 1946/2000. A new refutation of time. In Selected Non-Fictions. E. Weinberger (ed.), E. Allen, S.J. Levine & E. Weinberger (transl.), 317–333. New York NY: Viking Penguin. Bowie, A. 1997. From Romanticism to Critical Theory: The Philosophy of German Literary Theory. London: Routledge. Breuer, J., & Freud, S. 1893–1895/1957. Studies on hysteria. In The Standard Edition of the Psychological Works of Sigmund Freud, Vol. II. J. Strachey (ed. & transl.) London: Hogarth. Burke, K. 1969. A Grammar of Motives. Berkeley CA: University of California Press. Collingwood, R.G. 1938. The Principles of Art. London: OUP. Damasio, A.R. 1994. Descartes’ Error: Emotion, Reason and the Human Brain. New York NY: G.P. Putnam. Derrida, J. 1973. Speech and Phenomena, and Other Essays on Husserl’s Theory of Signs. Evanston IL: Northwestern University Press. Dickie, G. 1964. The myth of the aesthetic attitude. American Philosophical Quarterly 1: 56–65. Dilthey, W. 1887/1985. The imagination of the poet: Elements for a poetics. In Poetry and Experience. Vol. 5 of Selected Works. R.A. Makkreel & F. Rodi (eds), L. Agosta & R.A. Makkreel (transl.), 29–174. Princeton NJ: Princeton University Press. Ellis, R.D. 2005. Curious Emotions: Roots of Consciousness and Personality in Motivated Action. Amsterdam: John Benjamins. Fialho, O. 2007. Foregrounding and refamiliarization: Understanding readers’ response to literary texts. Language and Literature 16: 105–123. Freud, S. 1901/1960. The psychopathology of everyday life. In The Standard Edition of the Psychological Works of Sigmund Freud, Vol. VI. J. Strachey (ed. & transl.) London: Hogarth. Freund, E. 1987. The Return of the Reader: Reader Response Criticism. London: Methuen. Gendlin, E.T. 1962/1997. Experiencing and the Creation of Meaning. A Philosophical and Psychological Approach to the Subjective. Evanston IL: Northwestern University Press. Gendlin, E.T. 1991. Thinking beyond patterns: Body, language and situations. In The Presence of Feeling in Thought, B. den Ouden & M. Moen (eds), 25–151. Frankfurt: Peter Lang. Gendlin, E.T. 1992. The primacy of the body, not the primacy of perception: How the body knows the situation and philosophy. Man and World 2: 341–53.
Don Kuiken Gendlin, E.T. 1995. Crossing and dipping: Some terms for approaching the interface between natural understanding and logical formulation. Mind and Machines 5: 547–560. Gendlin, E.T. 1996. Focusing-oriented Psychotherapy: A Manual of the Experiential Method. New York NY: Guilford Press. Gendlin, E.T. 1997. A Process Model. New York NY: The Focusing Institute. Gendlin, E.T. 2004. The new phenomenology of carrying forward. Continental Philosophy Review 37: 127–151. Gibbs, W.R. & Kearney, L.R. 1994. When parting is such sweet sorrow: The comprehension and appreciation of oxymora. Journal of Psycholinguistic Research 23: 75–89. Gineste, M., Indurkhya, B., & Scart, V. 2000. Emergence of features in metaphor comprehension. Metaphor and Symbol 15: 117–135. Glucksberg, S. & Keysar, B. 1990. Understanding metaphorical comparisons: Beyond similarity. Psychological Review 97: 3–18. Glucksberg, S. & Haught, C. 2006. On the relation between metaphor and simile: When comparison fails. Mind & Language 21: 360–378. Grady, J., Oakley, T. & Coulson, S. 1999. Conceptual blending and metaphor. In Metaphor in Cognitive Linguistics. R.W. Gibbs, Jr. & G.J. Steen (eds), 101–124. Amsterdam: John Benjamins. Heidegger, M. 1927/1996. Being and Time. J. Stambaugh (transl.) Albany NY: State University of New York Press. Heidegger, M. 1953/2000. Introduction to Metaphysics. G. Fried & R. Polt (transl.) New Haven CT: Yale University Press. Iser, W. 1978. The Act of Reading: A Theory of Aesthetic Response. Baltimore MD: Johns Hopkins University Press. Jameson, F. 1991. Postmodernism, or, The Cultural Logic of Late Capitalism. Durham NC: Duke University Press. Kant, I. 1790/1951. Critique of Judgment. J.H. Bernard (transl.) New York NY: Hafner Press. Kennedy-Moore, E. & Watson, J.C. 1999. Expressing Emotion: Myths, Realities, and Therapeutic Strategies. New York NY: Guilford. Kuiken, D. 1998. Understanding the depth metaphor in aesthetic experience: Pressing the limits of psychological inquiry. In Toward a Psychology of Persons. W. Smythe (ed.), 101–117. Hillsdale NJ: Lawrence Erlbaum. Kuiken, D., Busink, R. Dukewich, T.L. & Gendlin, E.T. 1996. Individual differences in orienting activity mediate feeling realization in dreams: II. Evidence from concurrent reports of movement inhibition. Dreaming 6: 251–264. Kuiken, D., Busink, R. Miall, D.S. & Cey, R. Withdrawing to engage: How literary reading penetrates consciousness. Paper presented at the workshop “How Literature Enters Life II”, Utrecht, The Netherlands, June 26–28, 2003. Kuiken, D., Miall, D.S., & Aaftink, C. Reading that gestures toward what cannot be said: Sublime enthrallment and sublime disquietude. Paper presented at the conference of the International Society for the Empirical Study of Literature, Münich Germany, August 5–8, 2006. Kuiken, D., Phillips, L., Gregus, M., Miall, D.S., Verbitsky, M. & Tonkonogy, A. 2004. Locating self-modifying feelings within literary reading. Discourse Processes 38: 267–286. Lakoff, G. & Turner, M. 1989. More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago IL: Chicago University Press. Lyotard, J.F. 1988. The Differend: Phrases in Dispute. G. Van Den Abebeele (transl.) Minneapolis MN: University of Minnesota Press.
Chapter 4. A theory of expressive reading
Merleau-Ponty, M. 1945/1962. Phenomenology of Perception. Colin Smith (transl.) London: Routledge & Kegan Paul. Merleau-Ponty, M. 1964/1968. The Visible and the Invisible. Alphonso Lingis (transl.) Evanston IL: Northwestern University Press. Merleau-Ponty, M. 2002. Notes. In Husserl at the Limits of Phenomenology, Including Texts by Edmund Husserl, Maurice Merleau-Ponty, L. Lawlor & B. Bergo (eds). Evanston IL: Northwestern University Press. Miall, D.S. & Kuiken, D. 1994. Foregrounding, defamiliarization, and affect: Response to literary stories. Poetics 22: 389–407. Miall, D.S., & Kuiken, D. 1995. Feeling and the three phases of literary response. In Empirical Approaches to Literature, G. Rusch (ed.), 282–290. Siegen: Lumis-Publications. Miall, D.S. & Kuiken, D. 2002. A feeling for fiction: Becoming what we behold. Poetics 30: 221–241. Mukařovský, J. 1964/1932. Standard language and poetic language. In A Prague School Reader on Esthetics, Literary Structure, and Style, P.L. Garvin (ed.), 17–30. Washington DC: Georgetown University Press. Ogden, C.K. & Richards, I.A. 1923. The Meaning of Meaning. London: Routledge. Panksepp, J. 1998. Affective Neuroscience: The Foundations of Human and Animal Emotions. Oxford: OUP. Polanyi, M. 1964. Personal Knowledge: Towards a Post-Critical Philosophy. Chicago IL: University Of Chicago Press. Richards, I.A. 1930. Practical Criticism: A Study of Literary Judgment. New York NY: Harcourt Brace. Šklovskij, V. 1965/1917 Art as Technique. In Russian Formalism Criticism: Four Essays. L.T. Lemon, & M.J. Reis (eds. & transl.) Lincoln NB: University of Nebraska Press. Scruton, R. & Munro, T. 2003. Aesthetics. Encyclopedia Britannica. CD-Rom Ultimate Reference Suite (http://www.compilerpress.atfreeweb.com/Anno%20Scruton%20Aesthetics%20EB%2 02003%20a.htm#Introduction). Shen, Y. 1995. Constraints on directionality in poetic vs. non-poetic metaphors. Poetics 23: 255–74. Shen, Y. 1997. Cognitive constraints on poetic figures. Cognitive Linguistics 8: 33–71. Shen, Y. 1998. Zeugma: Prototypes, categories, and metaphors. Metaphor and Symbol 13: 31–47. Shen, Y. 2007. Foregrounding in poetic discourse: Between deviation and cognitive constraints. Language and Literature 16: 169–181. Shen, Y. & Balaban, N. 2005. Cognition and oxymoron. Paper presented at the International Workshop: Text and Cognition, May 15–18, Tel Aviv University. Sikora, S., Kuiken, D. & Miall, D.S. Enactment versus interpretation: A phenomenological study of readers’ responses to Coleridge’s ‘The Rime of the Ancient Mariner.’ Paper presented at the conference of the International Society for the Empirical Study of Literature, Utrecht, The Netherlands, 1998. Sopcák, P. 2007. Creation from nothing: A foregrounding study of James Joyce’s drafts for Ulysses. Language and Literature 16: 183–196. Stolnitz, J. 1960. Aesthetics and Philosophy of Art Criticism. Boston MA: Houghton Mifflin. Terada, R. 2001. Feeling in Theory: Emotion after the ‘Death of the Subject.’ Cambridge MA: Harvard University Press. Tourangeau, R. & Rips, L. 1991. Interpreting and evaluating metaphors. Journal of Memory and Language 30: 452–472.
Don Kuiken Ullman, S. 1957. Panchronistic tendencies in synaesthesia. In The Principles of Semantics. S. Ullman (ed.), 266–289. Oxford: Blackwell. van Peer, W. (1986) Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm. Wimsatt, W.K., Jr. 1954. The Verbal Icon: Studies in the Meaning of Poetry. Louisville KY: University of Kentucky Press. Wolpers, T. 1995. Recognizing and classifying literary motifs. In Thematics Reconsidered: Essays in Honor of Horst S. Daemmrich, F. Trommler (ed.), 33–70. Amsterdam: Rodopi. Zholkovsky, A. 1984. Themes and Texts: Toward a Poetics of Expressiveness, K. Parthé (ed.) Ithaca NY: Cornell University Press.
part ii
Psychology, Foregrounding and Literature
Psychology, foregrounding and literature Introduction The chapters in this section reflect seminal topics of Willie van Peer’s contribution to the empirical study of literature in many ways. They are dedicated to investigations on psycholinguistic aspects of textual comprehension, revealing that such interest shared by high-caliber scholars is not dissociated from a more socially relevant focus on the effects of narratives. From another perspective, these chapters reveal how science progresses even more solidly when interdisciplinary projects are undertaken. An example of a successful collaborative enterprise, Dixon and Bortolussi’s contribution stands as a most appropriate opening. The authors discuss how the intersection between empirical investigations of literary response and experimental science can be materialized by the use of experimental manipulation. In their view, it is an crucial approach to a deep causal insight into literary response. Two broad classes of experimental techniques are discussed: textual and extra-textual manipulation, and the advantages and liabilities of each technique are considered. The authors support the combination of different techniques in the pursuit of a better assessment of narrative. Their argument is illustrated by means of the analysis of three empirical studies, including the informative example of a textual manipulation used by van Peer (1990). The study of foregrounding is taken up by David Miall. His investigation of readers’ responses to a short story by Katherine Mansfield sheds light on the key role of foregrounded elements and the feelings aroused by them in reshaping readers’ perspective during the act of reading (see also Miall & Kuiken 2001). Miall provides empirical evidence indicating that not only foregrounding but also the degree of proximity to the inner world of main characters, as previously discussed by van Peer and Pander Maat (1996), are two aspects of narrative with important implications for the development of emotional models of perspective, a crucial issue in narrative theory (see van Peer & Chatman 2001). Chapter 7 also focuses on foregrounding. Here, Yeshayahu Shen discusses the intertwining aspects of this concept as used by discourse analysts and linguists, from one perspective, and literary theorists, from another. By analyzing several excerpts from various literary narratives, Shen identifies different levels of foregrounding and illustrates how such levels may combine to create what he calls “higher order foregrounding effects”. His article is a systematic attempt to relate
Directions in Empirical Literary Studies
these two different notions of foregrounding: the ‘linguistic’ and the ‘literary’, and it proposes how investigations of such relation can directly contribute to both areas of research. Shen brings the discussion on the effects of foregrounding to a different level and opens the possibility for a call for further empirical studies in order to verify the role of the foregrounded higher order effects he identifies in readers’ construction of an aesthetic experience. The interest in verifying the relation between narrative perspective and modes of participatory affective responses characterizes Chapter 8. Focusing specifically on narrative empathy and inter-group relations, László and Somogyvári provide evidence indicating that readers’ and characters’ group identities do not always influence narrative empathy and thus the impact of literary narratives. Some of the underlying questions posed not only by László and Somogyvári, but also shared by Miall, look at how narratives ultimately inform us about ourselves, our human nature and universal predicaments. The two following chapters by Mar et al. and Hakemulder shed further light on how literary narrative can benefit from new perspectives. Here we contemplate other possible selves and learn whether to accept or reject them. Taking from the theory that fiction is a kind of simulation that runs on minds (Oatley 1999), Mar et alii ask whether and how literature can change the reader. They argue that to read or watch such simulation is to pose social problems and practice on them. From an empirical perspective, they aim at showing how exposure to fiction may lead to important social outcomes as well as cause change in the sense of the self. It is these transformative powers of reading, they argue, that “guarantee the longevity and persistence of art across millennia of human civilization”. In the same line, Hakemulder replicates a previous study of his on readers identification with a story character (Hakemulder 2000) and shows how imagining ourselves in the shoes of characters might affect our beliefs about outgroups. His point is that if “reading literature stimulates us to reflect on what others think and feel, it may help us, in the end, to understand what they want. And maybe we discover we all want the same”. This is one of the fascinating insights cross-cultural studies can provide the realization that human beings, regardless of which culture they belong to, are more similar than what we can even possibly think of. The contributions to this section in one way or another look at how readers empathize with characters in narrative, how they take on their goals, feel their emotions, and draw upon the same social skills that enable them to understand others. Meaning in literature is always meaning for some specific person, from some specific point of view. Characters behave, but they also think and feel. They interpret their own behavior and that of others, and authors frequently give us to understand that these interpretations are partial, incomplete, biased, or distorted in some fashion. The experiences of reading vary widely, but as these accounts
Psychology, foregrounding and literature
as well as other studies indicate, readers often refer to the impact of reading on their self-development to explain why they read fiction. This makes the experience engaging, personally and emotionally relevant, being responsible for changes in how they think and feel. It is then possible to say that literary reading deepens understanding of self in the world and has implications for the way we live because of its transformative powers. According to Carroll (1995), that is exactly why art matters: because it imposes order, resolves uncertainties, providing models of reality that help organize human understanding. Hopefully, attending to evidence for such an effect of literature will help re-establish the function of the arts and the humanities in our school curricula, governmental budgeting, and in our society at large. Olivia Fialho Section Editor
References Carroll, J. 1995. Evolution and Literary Theory. Columbia MI: University of Missouri Press. Hakemulder, J. 2000. The Moral Laboratory. Experiments Examining the Effects of Reading Literature on Social Perception and Moral Self-Concept. Amsterdam: John Benjamins. Miall, D.S. & Kuiken, D. 2001. Shifting perspectives: Readers’ feelings and literary response. In New Perspectives on Narrative Perspective, W. van Peer & S. Chatman (eds), 289–301. Albany NY: State University of New York Press. Oatley, K. 1999. Why fiction may be twice as true as fact: Fiction as cognitive and emotional simulation. Review of General Psychology 3: 101–117. Van Peer, W. 1990. The measurement of metre. Poetics 19: 259–275. Van Peer, W. & Chatman, S. 2001. Introduction. In New Perspectives on Narrative Perspective, W. van Peer & S. Chatman (eds), 1–17. Albany NY: State University of New York Press. Van Peer, W. & Pander Maat, H. 1996. Perspectivation and sympathy: Effects of narrative point of view. In Empirical Approaches to Literature and Aesthetics, R.J. Kreuz & M.S. MacNealy (eds), 3–22. Norwood NJ: Ablex.
chapter 5
Textual and extra-textual manipulations in the empirical study of literary response Peter Dixon & Marisa Bortolussi Experimental manipulation is the best approach to a deep causal insight into literary response, and without it, firm conclusions are difficult. Here, we discuss two broad classes of experimental techniques. Textual manipulation involves exposing readers to different versions of a text that have been systematically varied by the experimenter and is ideally suited to studying the effects of textual features. Extra-textual manipulation involves presenting adjunct or surrounding material to the reader without changing the text per se and can provide insight into the role of the reading context and textual features under some circumstances. Examples of recent research will be scrutinized in light of this analysis. Keywords: literature, empirical studies, experimental methods, quasi-experiment, textual manipulation, causation
1. Introduction In this paper, we provide a framework for analyzing experimental investigations of literary response. By “experimental” we mean investigations that involve the deliberate manipulation of independent variables. We argue that such investigations provide the best means of developing an understanding of the underlying causes of readers’ responses to texts, and research without such manipulations is almost inevitably going to be limited in the nature of the inferences that can be confidently supported. However, an important distinction must be drawn between textual manipulations – that is, manipulations in which the text itself is changed – and extra-textual manipulations in which materials and circumstances surrounding the text are varied. Both of these types of manipulations can be found in research on literary response, but they have quite different strengths and weaknesses and need to be considered separately. After analyzing the basis of these two kinds of manipulations, we provide some examples from recent research that illustrate how
Peter Dixon & Marisa Bortolussi
experimental investigations incorporating these manipulations can support various causal inferences.
2. Variables in reading In analyzing and designing experiments, one must distinguish observable variables from unobservable ones. Observable variables are those that can be directly measured or manipulated, while unobservable variables are generally hypothetical theoretical entities in the reader. In turn, observable variables can be divided into textual variables, reader variables, and contextual variables. Textual variables refer to characteristics of the text that, at least in principle, can be objectively defined. It is important in this regard to distinguish textual variables from readers’ constructions. The former are properties of the text that are amenable to formal definition, independent of the reader, while the latter are the products of mental processes on the part of the reader and are potentially subjective and idiosyncratic. Despite our belief that textual features should be amenable to definition, we regard it as acceptable to investigate textual features for which it is not practical to provide an adequate definition given our current knowledge and technology. For example, one might be interested in whether the mood of a poem is humorous or serious. It is possible that a suitable semantic analysis could provide the basis of a workable definition of this feature. However, the amount of effort involved to do this adequately may be prohibitive. In such cases, we might use rating scales or consensus among experts as a working criterion for a given study. However, such an approach must be regarded as a proxy for more careful conceptual analysis that might be possible at some future point. The essential point, though, is that the process of ascertaining whether a given text has a particular feature needs to be well defined, so that it can be repeated by other researchers with other texts. Reader variables pertain to characteristics of the reader. There is a very wide range of reader characteristics that might be relevant to the investigation of literary response. Some examples of the variables that have been used include literary expertise, domain knowledge, cultural background, working memory capacity, and personality measures of various types. One reason that such variables have been so heavily researched is that they are assumed to be relatively stable dispositions in the reader. Thus, the nature of the causal relationship would seem to be fairly obvious: variations in the reader variables must be the cause of whatever variations in reactions are observed. However, it is quite possible that the reader characteristics of interest are correlated with other factors that are actually the causal agents. For example, the distinction between expert and novice readers of
Chapter 5. Textual and extra-textual manipulations
a text is generally confounded with a range of other variables. Experts are often more educated, more motivated, have greater verbal skills, have a greater breadth of experience beyond literature, and so on. Thus, finding a difference in reading process as a function of literary expertise does not definitively implicate expertise as the cause of those differences. Contextual variables are those that can be found in the situation in which the reading takes place. An important class of contextual variables includes those that affect reading goals. It is well established that the nature of the reading process and the information retained from a text varies with what readers intend to do with that information (e.g., Postman & Senders 1946 and Pritchart & Anderson 1977). In literary response studies, goals of the reader have often been studied by instructing the reader to engage in one form of processing or another. For example, Vipond & Hunt (1989) induced readers to search for the point of a story or to attend to the events and the characters; subsequently, readers´ reactions to the text were examined. Similarly, Cupchik, Oatley & Vorderer (1998) asked readers to feel sympathy for the protagonist as a spectator or to imagine what it is like to be the protagonist and then measured emotional reactions as a function of these goals. Contextual variables also include other factors, such as the environmental circumstances, the physical state of the reader, other ongoing tasks, and immediately preceding activities. Often the effects of contextual variables are mediated by unobservable variables in the reader. For example, instructions, which are an observable feature of the reading context, presumably affect the strategies and mental processes readers engage in. Although such mental processes are unobservable, it is reasonable in many cases to suppose there is a close relation between what wellintentioned subjects are asked to do in an experiment and what they actually do. Other types of mediating variables may be more subtle. For example, encountering the same text described as a story or a newspaper article may lead readers to activate a range of different expectations and schemas, as well as potentially quite different processing strategies (Zwaan 1994).
3. Causation and experimentation The central argument in favor of experimentation – that is, research involving manipulated variables – pertains to causal inferences. According to the arguments of Mill (1843), if two events co-occur, one of three causal relationships must exist: the first may cause the second; the second may cause the first; or some other, third event may have caused both. Equivalently, if variables A and B covary, the causal explanations consist of : (a) A ⇒ B, (b) B ⇒ A, or (c) C ⇒ A, B. Sometimes one
Peter Dixon & Marisa Bortolussi
of these possibilities can be eliminated on logical considerations. For example, it is not possible for a reader’s reaction to influence the objective features of a text rather than the other way around. However, possibility (c), third-variable causation, is difficult to eliminate a priori. The essential tool for doing so is an experiment: if A is a carefully manipulated independent variable in an experiment, there is only one possibility: A ⇒ B. Thus, the value of an experiment is that it makes inferences concerning causes much more straightforward. It is critical for causal inference that only the independent variable is manipulated, and no other variables change along with it. If this is not the case, the other, confounding variables provide alternative causal explanations. Eliminating possible confounds is thus a critical ingredient of good experimental design. A true experiment, with a manipulated independent variable, should be distinguished from a quasi-experiment in which a potentially causal variable is sampled rather than varied by the researcher. Quasi-experimental manipulations are a tempting avenue of investigation in the study of literary response. For example, if one is interested in whether a particular stylistic technique constitutes foregrounding, one may search for texts with and without that technique and assess whether there is some concomitant variation in readers’ reactions. Quasi-experiments lead to what we have referred to as “author confounds” (Bortolussi & Dixon 2003). Author confounds occur because texts are generally constructed by an author with the goal of producing intended reactions in the reader. As a consequence, the author may include any number of features in the text if he or she regards them as relevant to producing that reaction. For example, if the author intends to draw attention to and create appreciation of a particular scene in a story, he or she might use foregrounding techniques such as striking metaphors and novel allusions, both of which may be regarded (by the author) as helpful in producing the intended reaction. One can expect, then, that if a number of texts are sampled, these techniques will tend to co-occur whenever authors intended to draw attention to aspects of the story world. The net result is that when readers do attend to some aspect of the story, one cannot ascertain whether it is the metaphors or the novel allusions (or some other factor) that was responsible. A general strategy for addressing confoundings in quasi-experimental designs is to explicitly identify the potentially causal variables and then to control them. For example, one might search for texts that have striking metaphors but no novel allusions. If readers’ reactions persist in such cases, one may conclude that the use of novel allusions is not the sole cause of readers’ reaction and that the metaphors may have a role as well. Such control may also be exercised statistically by using the presence of allusions as a covariate in a suitable analysis. Within the bounds of the statistical model being applied, one may be able to conclude that the covariation between, say, the use of metaphor and the dependent variable cannot be attributed
Chapter 5. Textual and extra-textual manipulations
to the covariate, the use of novel allusions. The difficulty with this strategy is that each potentially confounding variable must be explicitly identified and controlled. While this is not necessarily an overwhelming burden in some domains, it is often difficult in the study of literary response because author confounds can produce a wide range correlated features in the text.
4. Manipulations Here we discuss textual and extra-textual manipulations, two broad classes of manipulations that can be used to investigate reading variables. Textual manipulations are ideally suited to investigating the effects of textual variables. While extra-textual manipulations are most relevant to contextual variables, they can also be used to understand reader and textual variables as well. Textual manipulations. Textual manipulations provide the most direct approach to eliminating author confounds in an experimental design. Based on a careful analysis of the text, one identifies a feature of interest and then creates different versions of the text that vary the presence, absence, or extent of that feature. The textual analysis is essential to ensure that only the target feature is manipulated and no others. This approach addresses the author confound because the researcher’s goal is to vary only a single feature to see whether that feature affects the reader. An author, on the other hand, generally is interested in manipulating as many features as possible to ensure that the text affects the reader. In a sense, an author’s goal is often to deliberately create author confounds, while the experimental researcher’s goal is to avoid them. Thus, a well-designed textual manipulation has the potential to provide definitive evidence concerning the causal relations between the target feature and readers’ reactions. The primary limitation of textual manipulations is precisely the essence of good experimentation: a manipulation should affect the variable of interest and no other. However, it is often difficult to ensure that no other variables have been changed when some aspect of the text is changed. We refer to such problems as “complexity confounds”: with complex literary texts, many features will be interrelated, and it may require some subtlety to control for unwanted effects of the manipulation. Thus, complexity confounds can easily occur if the researcher does not deeply understand the text and the relationships among its features. Importantly, though, complexity confounds are only relevant to causal inferences when the extraneous variables are related to the dependent variable. For example, one might be interested in how a character’s actions affect readers’ perception of that character’s personality traits, and a suitable manipulation might be to change the events of the story so that the character engages in different actions
Peter Dixon & Marisa Bortolussi
(cf. Bortolussi & Dixon 2003). Clearly, such a manipulation has the potential to affect the coherence of the plot. However, the confound with plot coherence could very well be unimportant if it has little effect on readers’ assessment of character personality. In general, the relevance of complexity confounds needs to be assessed with respect to the dependent variable being studied. Complexity confounds may be difficult to eliminate for some types of features or for some texts because the feature of interest is deeply entwined with other characteristics of the work. For example, in previous research we were interested in the role of character gender on readers’ interpretation of a character’s actions (Dixon & Bortolussi 1996). Gender is easy to manipulate in a text by simply changing the pronouns and names. However, doing so can easily affect many other relationships in the text since in our culture there are many gender-stereotyped behaviors and gender-based expectations. Readers might react very differently to a story concerning a female firefighter than they would to the same story with the same content concerning a male firefighter, for example, simply because a woman in that role is unexpected. In the Dixon & Bortolussi (1996) research, we addressed this issue by revising a number of minor aspects of the story to match readers’ gender-based expectations. However, this strategy is limited because some stories would require extensive revisions of this sort, and such large-scale changes would inevitably change the text in other ways. Two experimental techniques can also be used to minimize the impact of complexity confounds: combining textual and quasi-experimental manipulations and comparing textual manipulations with sham-manipulation controls. First, many of the inferential difficulties of textual manipulations can be resolved by crossing a textual manipulation with a quasi-experimental manipulation of the same variable. A recent unpublished study of our own makes this technique clear. We were interested in the effects of protagonist gender, and we started with two sets of texts, one that had male protagonists and one that had female protagonists. By itself, this contrast comprises a quasi-experimental manipulation of gender and is prone to a number of obvious author confounds. For example, the male and female characters in the selected stories may act and speak differently, have different reactions to events in the story, and so on, all of which could be a function of the intended effect of the character on the reader. Consequently, it would be difficult to conclude that any differing reactions of the reader to the male and female characters were due to character gender per se. However, we also created a modified version of each text in which the original male character was changed to a female character and vice versa. The shortcoming of this kind of manipulation alone is that it is prone to complexity confounds: changing a character’s gender could potentially interact with a wide range of other features of the story and the text, and it would be difficult to control for all of these possibilities. Both of these problems can be
Chapter 5. Textual and extra-textual manipulations
ameliorated by doing both the textual and the quasi-experimental manipulation at the same time. In this case, one would examine reactions to male characters (aggregated over the original and modified versions) and compare them to those for female characters (aggregated over the two versions as well). Under many circumstances, any overall difference can be attributed to protagonist gender, independent of author confounds and independent of complexity confounds. Second, the potential for complexity confounds can often be assessed with sham manipulations. A sham manipulation is a manipulation of the text that is comparable in important respects to the experimental manipulation but designed so that there is no theoretical reason to expect any effect on the reader. This provides a check on whether manipulation per se has an effect on the measures of interest. Such manipulations are commonly performed in animal research involving surgery. For example, if the experimental condition involves a surgical lesion in a particular part of the brain, sham-control animals will also undergo the identical surgical procedure but without the actual lesion. Thus, both experimental and control animals will have to recover from major brain surgery, and any differences between the groups cannot be attributed to the surgery per se. Textual manipulations are analogous to surgery: they might be thought of as a major intervention into a complex system of relationships in the text. Consequently, it is useful to ascertain whether any intervention at all would produce effects, and a shammanipulation control provides such a check. We used a sham-manipulation condition in a recent study of identification. In this research, we hypothesized that readers would identify with a character to a lesser extent if they were explicitly informed of the basis of the character’s motivations and attitudes. Consistent with this hypothesis, readers reported less identification when a paragraph describing that basis was added to the text (Bortolussi & Dixon 2003). However, one might be concerned that adding this (relatively unliterary) paragraph affected readers’ reaction to the story and to the character. To guard against this possibility, we compared the original story to one in which an equally unliterary paragraph was added but which did not provide information about the character’s attitudes. Because this sham manipulation did little to change reader’s identification with the character, we can be confident that adding a paragraph per se was not the source of the effects we observed in the experimental condition. Extra-textual manipulations. An extra-textual manipulation is one in which the text itself is left intact but surrounding or adjacent material varies across conditions. For example, one could provide different introductions to a target text that emphasize different aspects of the material, other similar or dissimilar texts might be provided as an experimental context, ancillary material might attribute the text to different sources, or different titles or prefaces might be provided. In all of these cases, the actual target text remains unchanged, so there is no possibility
Peter Dixon & Marisa Bortolussi
of complexity or author confounds. Instead, extra-textual manipulations have the potential to affect readers’ reactions to the text by influencing the reader’s attention, knowledge, expectation, or processing style. Generally, interpreting the effects of extra-textual manipulations involves hypothesizing mediating variables in the reader. For example, adding prefatory material to a text may influence how the reader interprets the source of the text and consequently what inferences readers might draw concerning the author’s intention. Because inferences regarding intentions are central to the reading process (Gibbs 1999), one can expect important differences in readers’ reactions. However, such an analysis requires assumptions concerning the processing of the author’s intention. Extra-textual information may also interact with reader variables. For example, an important class of extra-textual manipulations consists of additional material that is relevant to understanding the text. Such additions might include background knowledge concerning the setting of the story, information concerning the context in which the text was written, or discussions of the narrative techniques used in the text. Of course, such additional material would be particularly relevant if the reader needed that information to appreciate the text and if the reader was willing and able to use that information. Extra-textual manipulations can sometimes allow one to investigate the effect of textual features by affecting the manner in which textual features are interpreted. For example, the story “House Taken Over” by Julio Cortázar can be regarded as an allegory to the Peronist movement in Argentina, and readers without knowledge of that era would be unable to appreciate that aspect of the text. To assess the importance of the allegorical features of the story, one could design an experiment in which readers unfamiliar with the relevant Argentinian history are presented with an extra-textual description of the appropriate knowledge. To the extent that the allegory is central to readers’ reactions, one would expect to find an effect of the extra-textual manipulation, and the failure to find such an effect would suggest that the allegorical feature is not central to those reactions. This use of an extra-textual manipulation allows one to investigate the effects of textual features without risking complexity or author confounds. However, it depends heavily on theoretical assumptions concerning unobservable variables in the reader. In particular, one must assume that the extra-textual manipulation (in this case, background on Argentina) affects a mediating reader variable (knowledge of the Peronista period) that interacts with the feature of interest in the text (the allegory). While this seems plausible for the present example, such an analysis is not always possible, and extra-textual manipulations generally do not support direct insight into many effects of textual variables. To make the indirect nature of the inferences clear, an extra-textual manipulation can be contrasted with a textual manipulation with the same intent. For example, one could construct an alternative version of the Cortázar story in which
Chapter 5. Textual and extra-textual manipulations
cues to the allegorical nature of the text are systematically eliminated and compare reactions to this modified version to that to the original. Any differences between reactions to the text could then be attributed to the allegorical nature of the original. Such an inference requires no additional assumptions concerning readers’ knowledge or lack of knowledge and no assumptions about how that knowledge might be used in comprehending the import of the text. While such an approach is more direct than the corresponding extra-textual manipulation, it would be difficult to ensure that there were no complexity confounds. Both textual and extra-textual manipulations are subject to selection effects. Often one is interested in drawing inferences concerning readers’ reactions in general or the processing of literary texts quite broadly. However, empirical investigations in this area are rarely done with more than a handful of texts in any one instance. Thus, it is quite possible that the relationships that one observes are limited to just those materials and would not extend to other texts that are significantly different in some respect. This is particularly true if the materials are already deviant or unusual in some respect. Suppose, for example, one measured reading speed for a difficult text, such as Henry James’s The Europeans. Undergraduate subjects often find this text difficult to understand and have slow reading times. It would certainly be possible to improve readability by breaking up complex sentences into simpler ones, and one could anticipate that such a manipulation would lead to faster reading time. However, one could not conclude such a manipulation would always affect readability for any story, and material that is already easy to understand is unlikely to benefit. Instead, one should infer that the manipulation would be effective only for similarly difficult stories, or perhaps merely for difficult stories by Henry James. In other words, the materials used in an experiment should be regarded as representative of only limited population of comparable materials, and, strictly speaking, the conclusions one draws from an empirical study only really apply to that limited population.
5. Examples of experimental manipulations Here we analyze three studies that have used empirical data to draw causal inferences. First, we discuss a quasi-experimental manipulation of poetic style by Carminati, Stabler, Roberts & Fischer (2006); then, we describe an experimental textual manipulation of poetic meter by van Peer (1990); and finally, we summarize a recent unpublished study of our own on literary evaluation using an extra-textual manipulation. Quasi-experimental manipulation. Carminati et al. (2006) provide an example of a quasi-experimental manipulation. This study was designed to investigate readers’ sensitivity to poetry subgenres. Line-by-line reading times were collected for
Peter Dixon & Marisa Bortolussi
sections from an example of ottava rima (Byron’s serio-comic epic, Don Juan) and from an example of an elegy (Thomas Gray’s meditative and melancholy, Elegy Written in a Country Churchyard). Subjects read sections from one poem, and then, at some point, switched to sections from the other poem. The results indicated that according to a normalized reading-time measure, “reading efficiency,” subjects slowed when they switched from one poem to the other, suggesting that readers were sensitive to the differences in mood, style, and voice that characterized these subgenres. The authors wished to conclude that readers strategically adapt to fine-grained differences in poetry styles and that this adaptation is reflected in reading time. Because the difference in materials was pre-existing, the different conditions constitute a quasi-experiment rather than a true experiment. As a consequence, there are a variety of characteristics of the text that differed between the two conditions. For example, the poems concerned different topics, used different language styles, referred to different entities, and so on, and the reading time difference thus cannot be attributed unequivocally to any single variable. Of course, Carminati et al. (2006) were aware of this limitation and took some care to eliminate some of the obvious confounds. For example, sections of the poems were presented in different orders, so that the shift from one type of poetry to the other was uncorrelated with the particular phrasing and vocabulary found in any given line. In addition, the authors were concerned with the broad collection of textual features that characterize the two subgenres rather than with any single aspect. Nevertheless, author confounds could easily exist and undermine the interpretation of the results. For example, the conceptual content of the two differed in addition to the poetic subgenre, and the reading time effects might reflect such differences. Textual manipulation. Willie van Peer has produced many important contributions to the empirical study of literature using textual manipulations (e.g., van Peer 1990; van Peer & Maat 2001; van Peer, Hakemulder, & Zyngier 2007). An informative example of a textual manipulation can be found in van Peer (1990). In this study, van Peer examined the effect of metrical structure of a poem on the aesthetic and mnemonic reactions of readers. The original poem was written in a readily discernible iambic tetrameter, and a modified version was created that preserved the rhymes and the content but destroyed the meter. After reading the poem, subjects evaluated the poem on a range of aesthetic variables and performed a number of memory tests. The results indicated that the original, metrical version was rated more highly with respect to the aesthetic adjectives, particularly for “smooth” and “comprehensible.” Recognition memory for lines from the poem was also better for the original version. However, the manipulation had no effect on memory for the content. This study provides a clear demonstration that the metrical structure of the poem contributes to its aesthetic appreciation. The strength of this conclusion
Chapter 5. Textual and extra-textual manipulations
derives from the use of textual manipulation that allows one to rule out author confounds and related accounts. In particular, the design of the study involves a close control of the content, style, and form of the two versions. Thus, the comparison of the two conditions is not contaminated by distinct goals, intentions, or style of different authors. This observation is perhaps made more clear by considering how the question of interest might be addressed using a quasi-experiment rather than a textual experiment. In particular, the role of poetic meter could be addressed in by comparing readers’ reaction to a number of poems with clear metrical structure with that to poems that had no such structure. Even if readers preferred the metrical poems, it would be difficult to draw a strong conclusion because many other features of the texts differed besides metrical structure. For example, it seems quite plausible to suppose that authors who wrote metrical poems might have had very different aesthetic intentions than those who wrote non-metrical poems and that these differing intents may have been mirrored by many other characteristics of the poems besides metrical structure. Despite this strength, the van Peer (1990) study shares the problem that always accompanies textual manipulation, namely, that the changes in the text may have introduced other variations that were not intended. For example, the different vocabulary in the non-metrical version may not have been precisely comparable to that used in the original, and it is possible that the new word choices may not have been as apt or compelling. Such variation could conceivably have been a cause of the changes in the dependent variables. A possible check on this kind of alternative explanation would be to introduce a sham control as outlined above. For example, it might be possible to create a second modified version of the poem in which similar changes are made to the choice of words but that preserve the metrical structure. Thus, the sham control version would be, in some sense, just as modified as the nonmetrical version created by van Peer, but it would have the same metrical structure as the original. Consequently, if the effects observed in the van Peer study were due specifically to the manipulation of metrical structure, there should be no difference between the original and the sham-control version. If, on the other hand, effects are also observed with the sham control, it would undermine the conclusion that metrical structure alone was the causal variable. Extra-textual manipulation. Recently, we have completed a study in which we manipulated the attitude of the reader prior to reading a text by providing critical reviews of the work. The readers’ task was to read the first few pages of a novel and then assess the work on a number of evaluation scales spanning both subjective and objective reactions and assessments of both the story (i.e., the characters and
. We gratefully acknowledge the assistance of Paul Sopcak in developing the materials and collecting the data for this experiment.
Peter Dixon & Marisa Bortolussi
events of the story world) and the discourse (i.e., the language and style of narration). However, prior to reading the excerpt, they read a review of the book that was either positive or negative. The results are shown in Table 1 as a function of the four types of rating scales. Clearly, the reviews had an effect on readers’ evaluation of the work: ratings were higher following a positive review and lower following a negative review, and this effect was true of each type of rating. Table 1. Evaluations of novel beginnings as a function of type of evaluation and extra-textual critique (and standard error) Evaluation type Objective/Story Subjective/Story Objective/Discourse Subjective/Discourse
Negative critique
Positive critique
Positive – Negative
3.99 (0.13) 4.14 (0.13) 4.21 (0.13) 4.87 (0.13)
4.63 (0.13) 4.62 (0.13) 4.64 (0.13) 5.40 (0.13)
0.64 (0.19) 0.48 (0.19) 0.43 (0.19) 0.53 (0.19)
Several possible mechanisms might underlie this effect. One possibility is that the critique may have had an informational effect, so that the responses to the rating scales were some form of average of readers’ own reactions and the reactions of the reviewer in the critique. Alternatively, the critique may have had a biasing effect, so that the interpretation of the text as it is read is affected either positively or negatively by the nature of the preceding material. In this latter case, the extratextual manipulation would have been mediated by the online evaluation process or standards readers elect to use while reading. Further experimentation could disentangle these possibilities. For example, presenting the critiques after reading the text rather than before would have little effect on an informational averaging mechanism but would greatly attenuate biasing effects. Regardless of the particular mechanism involved, though, there is no ambiguity in these results concerning the nature of the causal relationship: the nature of the critique was a cause of the evaluations subjects produced.
6. Recommendations and conclusions The overarching thesis that we would like to emphasize is that experimental techniques are an essential tool to understanding causal relations. In particular, good experimental techniques allow one to quickly identify causal variables, while sound causal inferences are much more difficult using other approaches. This observation, of course, is the reason that experimentation is overwhelmingly common in science, and there is no reason why this lesson should not be taken to heart in the empirical investigation of literary response. In the present paper,
Chapter 5. Textual and extra-textual manipulations
we have contrasted textual and extra-textual manipulations and have argued that each is useful for different types of questions. In particular, textual manipulations are ideally suited to the investigation of the effects of textual variables. While extra-textual manipulations are central to understanding the role of context, they can, under some circumstances, provide insights into textual features as well. Moreover, there are costs and benefits to each kind of manipulation: while textual manipulations provide an ideal way of addressing author confounds, they are prone to complexity confounds; extra-textual manipulations are not susceptible to author or complexity confounds, but require strong assumptions concerning unobservable reader variables in order to draw inferences concerning the text. The best solution is thus to combine different types of experimental techniques to address a particular question: textual manipulations can be combined with counterbalanced quasi-experimental manipulations, and direct manipulations of the text can be contrasted with manipulations mediated by unobservable variables in the reader. In this way, the weaknesses of one technique are balanced by the strengths of another.
References Bortolussi, M. & Dixon, P. 2003. Psychonarratology: Foundations for the Empirical Study of Literary Response. Cambridge: CUP. Carminati, M.N., Stabler, J., Roberts A.M. & Fischer, M.H. 2006. Readers’ responses to sub-genre and rhyme in poetry. Poetics 34: 204–218. Cupchik, G.C., Oatley, K. & Vorderer, L. 1998. Emotional effects of reading excerpts from short stories by James Joyce. Poetics 25: 363–378. Dixon, P. & Bortolussi, M. 1996. Literary communication: Effects of reader-narrator co-operation. Poetics 23: 405–430. Gibbs, R.W. 1999. Intentions in the Experience of Meaning. Cambridge: CUP. Mill, J.S. 1843. System of Logic, Ratiocinative and Inductive, Being a Connected View of the Principles of Evidence and the Methods of Scientific Investigation. London: John W. Parker. Pichert, J.W. & Anderson, R.C. 1977. Taking different perspectives on a story. Journal of Educational Psychology 69: 309–315. Postman, L. & Senders, V.L. 1946. Incidental learning and generality of set. Journal of Experimental Psychology 36: 153–165. Van Peer, W. 1990. The measurement of metre. Poetics 19: 259–275. Van Peer, W., Hakemulder, J. & Zyngier, S. 2007. Lines on feeling: Foregrounding, aesthetics and meaning. Language and Literature 16: 197–213. Van Peer, W. & Maat, H.P. 2001. Narrative perspective and the interpretation of characters’ motives. Language and Literature 10: 229–241. Vipond, D. & Hunt, R.A. 1989. Literary processing and response as transaction: Evidence for the contribution of readers, texts, and situations. In Comprehension of Literary Discourse: Interdisciplinary Approaches, D. Meutsch & R. Viehoff (eds), 155–174. Berlin: de Gruyter. Zwaan, R.A. 1994. Effect of genre expectations on text comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition 20: 920–933.
chapter 6
Foregrounding and feeling in response to narrative David S. Miall This chapter provides a framework for considering some of the determinants and implications of readers’ feelings in response to a story by Katherine Mansfield, previously featured in an empirical study of foregrounding (Miall & Kuiken 1994). In additional analyses I focus on feelings that distinguish the story’s episodic structure, its narrative perspective, and phonetic patterns, and consider their effectiveness in post-predicting readers’ data from the 1994 study. Finally, I examine those processes of feeling that may be distinctive to literary response (Miall & Kuiken 2001), and situate the discussion in relation to a theory of readers’ feelings offered by Oatley (2002). Keywords: foregrounding, laws of emotion, narrative, episodes, perspective, phonetics
1. Response to fiction: Generalizing the model A modernist short story, such as those by Katherine Mansfield, typically focuses on a single consciousness and leads to a culminating insight. In our study of foregrounding (Miall & Kuiken 1994) that focused in part on Mansfield’s “The Wrong House” (Mansfield 1945/1919), we introduced a theoretical approach based on the premises of Romantic theory, Russian Formalism, the Czech critic Mukařovský, & van Peer’s (1986) groundbreaking empirical study, in which stylistic features are thought to defamiliarize the reader. We were able to show that passages high in foregrounding not only were found striking, but took relatively longer to read compared with low-foregrounded passages, and also evoked feeling. We were not concerned in that study with the larger, experiential aspects of the story, such as where insight is generated or how readers might construe it. In turning to consider the issue of readers’ understanding of the story as a whole, however, I ask a more
David S. Miall
problematic question: what aspects of a text evoke reader’s attention and interest, and how do they interact with the psychological processes required to reach an understanding of it? In discussing the modernist period, Clare Hanson (1985) has shown the presence of two classes of short story, one based around a particular situation or short narrative episode, the second largely devoid of plot and focusing on a moment of truth in experience, what James Joyce termed an “epiphany” (6–7). Mansfield’s “The Wrong House,” like much of her other short fiction, seems predominantly to belong in the second category. As Hanson also points out, this kind of story in the hands of writers such as Woolf, Joyce, and Mansfield, is notable for the development of free indirect discourse where, in Hanson’s phrase, “the voice of the narrator is modulated so that it appears to merge with that of a character of the fiction” (56). The modernist story is thus inclined to philosophical relativism and “best able to express a fragmented sensibility” (57). Thus, in “The Wrong House,” the most striking parts of the story are mediated through the consciousness of the old woman, and the culminating moment of the story appears to be an epiphany on her part – a markedly negative one (Joyce, however, did not require an epiphany to be a positive experience: cf. the endings of “Araby,” or “A Painful Case”). In describing her state of shock as the undertakers depart – “she thought of nothing; she did not even think of what had happened” – Mansfield portrays a fragmented consciousness, but also conveys obliquely the old woman’s horror at this sudden intrusion of death. For the reader, is the insight offered by the story specific to that character at that particular moment, or does it have a general symbolic power? In considering how readers construe a literary text, one place to focus is the episode. We have recently begun to consider empirical evidence for the reality of episodes as a unit for the analysis of readers’ responses (Miall 2006: Chapter 8). In the case of Mansfield’s story some support is available from our earlier analysis of the “phases” of response, as we have called them (Miall & Kuiken 2001). In our conception of the phase, we suggested that a peak in foregrounding would be defamiliarizing for readers. Since existing schemata are inadequate for understanding at such moments, feeling arising from the moment of foregrounding would become the vehicle for seeking a new understanding; this might take some time to come into place. If this hypothesis is correct, then near the beginning of such a phase, stronger feeling in a given segment would be marked by a lengthened reading time, since feeling here would signal greater uncertainty and the search for meaning. At the end of the phase, on the other hand, feeling is the vehicle for new understanding, thus segments with stronger feeling would direct reading, making . The text of Mansfield’s story is available at: http://www.ualberta.ca/~dmiall/reading/ Mansfield.htm.
Chapter 6. Foregrounding and feeling in response to narrative
it more certain; here, stronger feeling should be marked by a shorter reading time. To investigate this possibility, we examined correlations of feeling ratings per segment and reading times within a moving window of 6 segments using data derived from our 1994 study (where the story was divided into 84 segments, roughly equivalent to a sentence). For all three stories in that study, a pattern of alternating positive and negative correlations emerged, corresponding to the opening and closing of successive phases. Phases in the Mansfield story are shown in Figure 1 (the curve has been smoothed by averaging each data point with its neighbor). The curve for the time/ feeling correlations provides some support for the reality of the episodic structure of the story. In episode 1 (segments 1–12) the curve remains low, suggesting that the scene portrayed is readily identifiable by readers, since stronger feeling correlates with shorter reading times. Near the beginning of the second episode (13–27), however, the atmosphere of the dusk around line 20, “dusk came floating into the room,” corresponds to a peak in positive correlations, which is followed by a drop back to the negative as this information is assimilated. Similarly, episode 3 (28–64) shows an early positive peak as the funeral appears, dropping as the old woman judges the undertakers and assumes they are passing by; then, with the knock on the door the curve rises markedly again to the positive and remains high until the end of the episode (this longer episode, although united by action, space, and time, could be divided into two parts, given the marked shift in focus at segment 40). The last episode also shows an early positive peak followed by a fall to
Time-feeling correlations
0.500 0.250 0.000 –0.250 –0.500 –0.750 0
10
20
30
40 50 Segments
60
70
80
Figure 1. Correlations of reading time and feeling ratings within a moving window of 6 segments.
David S. Miall
the negative. Thus, apart from the first episode, we see each episode characterized by an early positive peak in time/affect correlations, indicating readers’ need to understand and assimilate new information; as shown by negative correlations, this understanding then comes into place towards the end of the episode, except for the third climactic episode, where the old woman’s response to the funeral creates a crisis of understanding, not only for the woman but perhaps also for the reader. In developing the notion of phases, we chiefly had in mind the role of foregrounding in creating uncertainty in the reader, which would be followed by a positive peak in the time/feeling correlation. In the light of the present analysis, however, this now seems to constitute only one influence on the constructive role of readers’ feelings and their variable reflection in reading times. The second influence is the extra processing requirements involved in instantiating a new episode. It seems likely that the feeling evoked by foregrounding will play a role in this process, but other influences deriving from narrative features must also play a role. In the next section some additional indicators for the analysis of both narrative and stylistic features are outlined.
2. Characterizing narrative segments In our earlier work on narrative aspects of response (Miall & Kuiken 2001), we examined the situation model theory of Zwaan et al. (1995), and concluded that two other aspects of narrative specific to literary texts should be addressed if we were to predict more effectively the processing time required during reading. First, segments should be indexed for foregrounding (Miall & Kuiken 1999); second, proximity to the inner world of the main character should be indexed (cf. van Peer & Pander Maat 1996). The more the narrative invites the reader to experience the internal thoughts or feelings of the main character, the longer the processing time required of the reader. Thus we developed a four-point perspective scale (see Miall & Kuiken 2001: Table 1), which was found to have some value in predicting reading times in a regression analysis that also included situation model factors and foregrounding. For the present study, however, I re-examined the perspective scale and noticed that it conflated two different components, perspective and feeling. Thus I created two separate indices that capture potentially different aspects of readers’ responses to a main character. A revised Perspective scale indicates the degree to which the view of a character is external or internal, and within this how far the reader is enabled to share the character’s own perceptions and motives. A segment by segment score on a scale of 0 to 4 is based on these criteria:
Chapter 6. Foregrounding and feeling in response to narrative
0. No reference to character, or reference by a third person only. 1. Character’s situation or appearance is described from an external, observer perspective, including affectively neutral indirect discourse. 2. Direct discourse or behavior of the character, where this seems unmarked by explicit feeling. 3. Character’s cognitions: perceptions, thoughts, etc., where feeling is at a low level; direct discourse that invites the reader’s involvement in character’s perspective. 4. Feelings of character, where these appear to raise motivational issues for character; motives and emotions; most passages of free indirect discourse; direct discourse only if it enables the reader to share character’s emotions or motives. It will be noticed that a character’s feeling is now only one of the possible components contributing to perspective. For example, free indirect discourse creates perhaps the most intimate proximity to a character. Thus, segment 19, “Only three? It seemed dusk already,” is rated a 4, although it implies little in the way of feeling. Segment 35, “What horrible-looking men, too! laughing and joking,” is strongly marked by feeling, but is also free indirect discourse. The segments are equivalent, however, in providing the most direct representation possible of the character’s consciousness. A segment that reports perception or thought rather than representing it in free indirect discourse is rated a 3, such as segment 47, “She saw the man withdraw a step, and again – that puzzled glance at the blinds.” The Perspective scale can be considered a predictor of readers’ responses obtained during the 1994 study. It appears to be factor in the judgments readers made in their ratings for feeling and strikingness. A regression analysis with segment position, length of segment, foregrounding, and perspective as predictors of mean feeling ratings is significant overall, F(4,81) = 7.691, and shows each of the predictor variables except segment position to be significant. The same analysis with strikingness ratings as the dependent variable is also significant, F(4,81) = 12.177. On the other hand, perspective has no significant relation to mean reading times per segment. However, an important component of Perspective appears to be foregrounding, since the two indexes correlate highly, r(84) = .302, p < .01. The primary component in this relation is the subindex for grammatical foregrounding, r(84) = .298, p < .01. The second index is for feeling. This scale seeks to capture the state of feeling in the main character in each segment, as follows: 0. No reference to main character. 1. Character’s situation, actions, or appearance are described, but from an external perspective; no feeling.
David S. Miall
2. Through action, discourse, or appearance an underlying feeling or mood of the character is implicitly represented or connoted. 3. Through action, discourse, or appearance an emotion is represented or connoted. 4. Feeling of character is described explicitly, or shown by free indirect discourse, or strongly indicated by character’s expression. As the description for the highest score shows, free indirect discourse is one possible marker of strong feeling. However, direct discourse or description can also indicate strong feeling, as in segment 44, “ ‘No!’ she groaned.” A less strong moment of feeling may also be connoted by free indirect discourse, as at segment 19: “Only three? It seemed dusk already,” which is scored a 3. However, more general moods without immediate implications for the story are scored a 2, such as segment 10, “Nothing. It was a habit. She was always sighing.” Again, the scale appears to have predictive value. A regression analysis, now with the feeling scale along with segment position, segment length, and foregrounding, show all to be significant predictors of readers’ feeling ratings, F(4,84) = 8.339; a similar analysis with strikingness ratings is also significant, F(4,84) = 12.611. Again, there is no relationship with reading times. As with perspective, significant correlations occur with grammatical foregrounding, r(84) = .283, p = .01, and with overall foregrounding, r(84) = .312, p < .01. Thus, while the predicted relationship between the two scales and reading times did not occur, it is evident that at the segment level an important role in readers’ judgments of the effect of the story was played by shifts in perspective and variations in access to the feelings of the main character. A third method for characterizing the story is to consider it at the level of sound. It is evident from Mansfield’s own comments on her writing that she considered this level of major importance for her craft as a writer. In a letter to Richard Murry dated January 17, 1921 writing about “Miss Brill,” a story published two years before “The Wrong House,” she said: It’s a very queer thing how craft comes into writing. I mean down to details. Par example. In Miss Brill I chose not only the length of every sentence – I chose the rise and fall of every paragraph to fit her – and to fit her on that day at that very moment. After I’d written it I read it aloud – numbers of times – just as one would play over a musical composition, trying to get it nearer and nearer to the expression of Miss Brill – until it fitted her. (Mansfield 1984: IV, 165)
Mansfield is probably referring here to several components of sound, including the rhythm of each sentence and the choice of words and word patterns based on their phonetic coloration. For “The Wrong House” I carried out an analysis of phonemic patterns, enabling each segment of the story to be characterized in
Chapter 6. Foregrounding and feeling in response to narrative
terms of its overall pattern of vowels and consonants. This analysis depends first on producing a transcript of the words of the story into phonemes, based on the 20 vowels and 24 consonants in standard English. The vowels are then each given a numeric weight according to two dispositions, a front-back ordering (based on the position at which the vowel is produced in the oral tract) and a high-low ordering. The consonants are weighted similarly according to a high-low and a soft-hard ordering (see Miall 2001). A story segment can then be given an overall weighting on each measure by cumulating the weights for individual phonemes. In addition, two other phonetic measures are provided for each segment: vowel shift (the vowel lengthening that occurs before voiced stops and fricatives), and absolute vowel length, that allows for diphthongs. The phonetic measures were then tested as predictors of the data provided by readers in our 1994 study. A series of regression analyses was carried out that included the six phonetic measures and the number of syllables per segment as predictor variables. The overall regression model was very significant in the case of reading times, F(7, 78) = 34.45, p < .0001; not significant for mean feeling ratings, F(7, 78) = 1.66; but significant for strikingness ratings, F(7.78) = 4.21, p < .001. Each variable was then examined separately (excluding the feeling ratings model). The partial correlations with the dependent variables are shown in Table 1. Table 1. Phonetic variables in Mansfield’s story as a predictor of reading variables Predictor variable Syllables Vowel shift Vowel length Vowels: front-back Vowels: high-low Consonants: soft-hard Consonants: front back
Reading time .595** .022 .113 − .297* − .011 − .283* .135
Strikingness .022 .306* .148 − .305* − .015 .033 .055
*p < .01, **p < .001 (two-tailed).
These findings show that reading times were strongly influenced overall by phonetic variations: the longer reading times are associated with back vowels and hard consonants. For example, the lowest weighted line showing a preponderance of back vowels is 30, “Good gracious! It was a funeral.” One of the lowest-weighted segments for hard consonants is 54, “She was shutting the door again when he fished out of the tail of his coat a black, brass-bound notebook and swiftly opened it” (this is due particularly to the numerous plosives, /t/, /d/, and /b/). In addition, the segments found to be the most striking are those with the larger proportion of back vowels.
David S. Miall
Shifts in phonetic tone are also apparent in the story as a whole. The mean weights of the phonetic measures for each episode show a marked shift: most notably, the soft-hard consonant weights in the crucial third episode shift from relatively soft in the first part of the episode to predominantly hard in the second part. These differences also seem consistent with the tenor of the story, and in keeping with Mansfield’s declared effort to discriminate the phases of her characters’ experience at the level of sound.
3. The modifying power of feeling In this last Section I offer a sketch of the reading process overall, based on the premise that feeling is the primary vehicle by which a reader comes to understand Mansfield’s story. In Miall & Kuiken (2002) different types of feeling are examined, including such feelings as suspense or curiosity, empathy with a character, or pleasure in the effectiveness of a metaphor. But we proposed that literary texts are distinctive for going beyond such contributory feelings, important though these are. In particular, a text evokes the reader’s feelings in order to modify both understanding and feelings themselves. I will demonstrate this with examples from my own reading of Mansfield’s story. The second segment of the story provides an example of the feelings evoked as I read the story: “Like an old song, like a song that she had sung so often that only to breathe was to sing it, she murmured the knitting pattern.” The ritualized, unthinking repetition of the knitting pattern here is situated for me in a similar experience, that of repeating the liturgy in the church I attended as a child, where, like the compacted phrase “woolinfrontoftheneedle,” I also sang words that had run together and had become devoid of meaning. The feeling thus brings together two apparently unrelated domains, knitting and church services. The feeling is self-referential, bringing to mind an extensive set of memories from childhood (their relevance to understanding this story, however, ranges from the highly apt to the irrelevant), but perhaps, more important, a stance of the self towards language that seems familiar and still a potential source of meaning (as I hear political slogans, for example). The feeling can also be considered anticipatory, since it is now one perspective on the old woman that I will hold in place in order to assess subsequent information. In the next few segments, for example, it suggests to me that the motive for the vests being knitted has long since become vestigial, and if there was a time when “a photograph of repulsive little black objects” aroused her compassion, that time has long passed. More generally, the feeling (for me) points to the episode further on in the story when the funeral appears, and I sense the routine nature of the undertakers’ activity, suggested in the description of them “laughing and joking.”
Chapter 6. Foregrounding and feeling in response to narrative
The feeling associated with a ritual that has become almost devoid of meaning, which is probably distinctive to my reading, has created an ad hoc category, in the terms of Glucksberg & Keysar (1990). This serves to organize other features of my response to the story, an organization that is both hierarchical and horizontal. Hierarchically, the feeling helps to place the old woman for me in a particularized social setting with carefully maintained class divisions (she has a maid; she has the leisure to knit vests for charity), with a history that is intimated, “She was always sighing,” and with limitations that her murmur and her sighing intimate. Horizontally, the feeling enables me to discriminate (rapidly and largely unconsciously, until I stop to consider it here) the various types of evidence for the old woman’s position. The feeling seems to indicate that her murmur and sighing are more indicative evidence of what she represents than the mission vests she is knitting, and that in her world the vests are more important than the “little black objects” for whom they are intended. Another process attributed to feeling (Miall & Kuiken 2002) is catharsis. In our reinterpretation of this concept, we proposed that the modification of hubris by pity and fear (the classical example originating with Aristotle’s discussion of Oedipus Rex) is one particular form of a more general pattern of response, in which one feeling modifies or recontextualizes another. In the case of Oedipus, we suggested, close attention to the play will show that fear first modifies hubris; fear is then itself modified by pity. A comparable process, albeit on a less universal scale, is apparent in “The Wrong House.” The succession of major feelings evoked by the story (at least in my reading) is significant in part because on several occasions a new feeling modifies an existing one. This can be seen as a special case of class inclusion. I will briefly describe two examples. In the second episode, time appears as a significant factor in the old woman’s consciousness: “Only three? It seemed dusk already.” This begins (for me) to recontextualize in a minor way the effect of ritual put in place at segment 2: the murmuring and sighing now seems to be a way of rendering time itself devoid of meaning. Why, I wonder, would the old woman be concerned about time, when her life seems devoted solely to passing time in a ritual, unthinking manner? It is the next segment, however, that places the ritual in a new perspective: “dusk came floating into the room, heavy, powdery dusk settling on the furniture, filming over the mirror.” Here, probably like many readers before me, I read dusk but also think of dust. The “powdery dusk” settling on everything is an evocative symbol for death, suggesting both the traditional view of dying as the valley of shadow, and the funeral service that speaks of “dust to dust.” Now the feeling of ritual, which was at first little more than a half-forgotten memory of childhood church services, suddenly takes on a sharper focus, giving to those embedded in such ritual an odor of imminent death. Ritual is recontextualized by the feelings of dusk/dust. For the first time I have a sense of something ominous to which her ritual behavior
David S. Miall
has made her vulnerable. The sense of catharsis, then, occurs when a new feeling recontextualizes an existing feeling. A major turning point in feeling, however, occurs as the old woman hears the knock on the door. My sensibilities, such as they are, enable me to see the old woman as little more than a comic figure as she realizes the funeral procession has stopped outside. Even her excessive fright seems comic: “Her old heart leaped like a fish.” My indifference to her predicament is halted abruptly, however, at segment 45. She knows the undertaker is about to knock, and she groans “No.” Then we read, “But yes, the blow fell, and for the moment it struck her down.” Suddenly, the prospect of the funeral strikes home: I feel the blank terror on her behalf. This is because Mansfield does not mention the knock on the door: rather, she renders its effect subjectively as “the blow”; and subjectively this is far worse than a mere knocking could be. In it echoes the tradition of the fatal knock on the door, from Macbeth, guilty over the murder of Duncan, to Don Giovanni hearing the arrival of the stone guest. Moreover, this blow “struck her down,” a phrase that (in my reading) is clearly not a literal falling, but a much worse one, a moral defeat; the blow that renders helpless. Feeling here, then, undertakes a sudden reversal: my indifference to the old woman’s fear of the funeral procession stopped outside her door is changed to a feeling that comes much closer to my own sense of mortality. In the language of catharsis, we might say that my previous indifference is a form of hubris that is unexpectedly recontextualized by fear. It would be appropriate to consider, finally, what status these feelings might have. I have argued specifically for several forms of modification due to feeling, from generalization, through class-inclusion, to catharsis, and I have suggested that these processes of feeling may be distinctive to the literary domain. The account of emotions in literature offered by Oatley (2002) focuses principally on emotion as a simulation (continuing a theme developed by Oatley in several previous publications), but adds to this some proposals drawn from ancient Indian poetics. It seems worth indicating how far the proposals discussed here overlap with those of Oatley and to what extent they differ. In introducing the topic of emotional response to literature, it seems clear that Oatley has in mind a modifying process comparable to the one I have presented here. As we read about the vicissitudes of characters’ lives we identify with them, he remarks, and experience emotions in sympathy – moments captured by higher scores on the perspective or feeling index, or both; and as our correlation analysis showed, this points to the role of foregrounding in evoking them, and readers’ experience of strikingness as they read such passages. For Oatley, to read fiction is to run “a kind of simulation, one that runs on minds rather than on computers.” While the simulation is running, emotions enable “identification with a protagonist, sympathy for story characters, and activation of emotional autobiographical memories that resonate with story themes” (41).
Chapter 6. Foregrounding and feeling in response to narrative
In Oatley’s view, this is what Aristotle meant by mimesis: not copying or imitating, but simulating in the sense that the reader creates a model of the fictional events that runs in parallel to the real world (48). Thus Oatley prefers to say that as readers we enact or perform our reading of fiction (50). This account suggests that our concepts and feelings do not remain unchanged during reading, and Oatley points to this possibility. During reading, “our emotions may be transformed by having them deepened or understood better, and they may be extended towards people of kinds for whom we might previously have felt nothing” (43). Similarly, reading recruits emotions and memories, and these “start trains of thought that readers would not otherwise have had . . . They prompt new connections within the self, and they elaborate meanings, which can be built into our mental structure as parts of ourselves” (55). In discussing my responses to the Mansfield story, I indicated similar moments of transformation: the ad hoc category of ritualized behaviors is a new train of thought; the feeling associated with ritual behaviors is then itself altered by its recontextualization when dusk/dust begins to settle heavily within the old woman’s room, with its connotations of death. My account attempts to provide a detailed mechanism for how such modifications are effected: through the response to foregrounding, generalization, ad hoc categories, and a new model of catharsis. The principal difficulty in Oatley’s account is that the status of the emotions aroused by fiction is left ambiguous. Do they really help to illuminate the self (cf. van Peer 1997: 220)? Are the emotions of fiction as real as those of daily life? Oatley’s appeal to Goffman seems to propose this, by suggesting that transformations during fiction are comparable to the transformative encounters with other people that Goffman has described (cited 42). On the other hand, by appealing to the Indian concept of rasas (52) – emotions unique to the aesthetic experience, according to Indian poetic theory, such as Delight, Anger, or Heroism – Oatley accepts that a fictional emotion is not the same as a real emotion. He adds that in fiction we have an element of choice over whether to engage with the offered emotion; and for modification to be possible both a certain distance from the emotion and space for reflection are necessary (64). Oatley attempts to bridge these two views by suggesting that they constitute two phases of the response process: We cannot always be moved and think about something in an observational way at the same time. What we can more often do is to move in and out along the continuum of emotional distance, be fully engaged emotionally at one moment, and then in the glow of that emotion, think about the experience in a more distanced way (64).
This view challenges, appropriately I believe, the notion that our usual mode of fictional reading is that of complete absorption, or a trance-like state (e.g., Nell 1988; Birkerts 1994), and it begins to suggest an answer to the problem we have
David S. Miall
inherited from Coleridge over whether reading requires the “willing suspension of disbelief ” (Coleridge 1983: II, 6; cf. Gerrig 1993: 17). However, there is a residual problem, which Oatley does not entirely resolve, that stems from overlooking a central property of feeling. In brief, a feeling considered in itself contains no information about whether its cause is fictional or real. Nor is a feeling temporally marked: as Coleridge put it, “All intense passions have faith in their own eternity” (Coleridge 1957–2002: III, 4056). In emotion theory, these aspects of feeling are indicated the most clearly in several of Frijda’s (2007) “laws of emotion.” First, the Law of Apparent Reality proposes that emotion is only elicited by what is taken to be real; second, the preoccupation of the mind by a given emotion is expressed by the Law of Closure – an emotion takes over the action system, hence the apparently involuntary nature of emotion; third, the Law of Conservation of Emotional Momentum states that unless counteracted, an emotional event retains the power to evoke emotion indefinitely. It is these aspects of emotion that the Indian concept of rasa also suggests in its own way: as an aesthetic emotion, a rasa has no relation to time but constitutes a perception of an eternal state of being. At the same time, a rasa is also considered separate from everyday emotion, being an emotion that is experienced only in the context of art (Gnoli 1968). However, when experienced it is felt to be real in a way that makes ordinary life seem illusory, thus rasa does not involve suspension of disbelief so much as suspension of what we take to be reality; thus whether a rasa is real or not is not in question. The reading of fiction depends upon both feelings and cognitions. It is a dialectical process, bringing into play both states of feeling that are indifferent to time and reality, and transient details of plot, character, situation, time, and space. However, the powers we have attributed to feeling – generalization through cross-domain, anticipatory, and self-referential processes, the creation of ad hoc categories with distinctive hierarchical and horizontal properties, and the recontextualization of one feeling by another in catharsis – each seems to call on the Laws of Emotion, i.e., the indifference of emotion to time or reality, while literary fiction is unique in challenging emotion on precisely these grounds. Thus, while my sense of ritual in the opening scene of Mansfield’s story is augmented as an ad hoc category of some power, with a reality of its own and a value that transcends time, a subsequent episode of the story alerts me to its limitations: the invasion of the dusk/dust shows me that the feeling renders me liable to a misreading of the evidence, not only in the story itself but potentially in my own experience. The heaviness of the dusk and its relation to the time of day has modified my earlier understanding, and at this point a sense of something like a rasa has come into place, perhaps the rasa of Fear. Now, too, I can retrospectively see this larger feeling underlying the evidence for the old woman’s behavior as she knits and sighs, leading me to modify my
Chapter 6. Foregrounding and feeling in response to narrative
understanding of what this means (for me as reader, that is, not yet for the old woman herself – that lies a little further off as the funeral pauses outside her door). This new feeling itself, the rasa of Fear, is timeless and indifferent to reality, and this power has enabled it to recontextualize and modify the earlier feeling. This one, though, will be modified in turn by others that occur later in the story. In this way, while a particular feeling does not give up its timelessness and indifference to reality, as a reader the fiction places me in a dialectical relation to it, able to be aware simultaneously of its intrinsic or prototypical nature and of its limitations as an evaluation of the situation now evolving in the story. The concept of rasa captures the first aspect, and points to its pervasive nature as it comes to imbue and color all aspects of a given fiction (hence Frijda’s Laws); but it fails to account for the dynamic process by which the rasa emerges, that is, how feelings gain their generalizing power, and the power to recontextualize and modify other feelings. The reading of fiction remains a paradoxical experience, half in our real world of everyday life and half outside it. But this paradox seems due directly to the double role of feelings, being in themselves timeless and indifferent to reality, yet within a literary context capable of being the focus of critical awareness and subject to the modifying power of new feelings.
References Birkerts, S. 1994. The Gutenberg Elegies: The Fate of Reading in an Electronic Age. New York NY: Fawcett Columbine. Coleridge, S.T. 1983. Biographia Literaria, Vols. 2, J. Engell & W.J. Bate (eds). London: Routledge. Coleridge, S.T. 1957–2002. The Notebooks. Vols. 5, K. Coburn (ed.). London: Routledge. Frijda, N. 2007. The Laws of Emotion. Mahwah NY: Lawrence Erlbaum. Gerrig, R.J. 1993. Experiencing Narrative Worlds. New Haven CT: Yale University Press. Glucksberg, S. & Keysar, B. 1990. Understanding metaphorical comparisons: Beyond similarity. Psychological Review 97: 3–18. Gnoli, R. 1968. The Aesthetic Experience According to Abhinavagupta, 2nd edn [Chowkhamba Sanskrit Series Office, Vol. LXII]. India, Varanasi-1. Hanson, C. 1985. Short Stories and Short Fictions, 1880–1980. Basingstoke: Macmillan. Mansfield, K. 1945. The Wrong House. In Collected Stories, 675–678. London: Constable. (Story first published in 1919). Mansfield, K. 1984. The Collected Letters, Vols. 4, V. O‘Sullivan & M. Scott (eds). Oxford: Clarendon Press. Miall, D.S. 2001. Sounds of contrast: An empirical approach to phonemic iconicity. Poetics 29: 55–70. Miall, D.S. 2004. Episode structures in literary narratives. Journal of Literary Semantics 33: 111–129. Miall, D.S. 2006. Literary Reading: Empirical and Theoretical Studies. Frankfurt: Peter Lang.
David S. Miall Miall, D.S. & Kuiken, D. 1994. Foregrounding, defamiliarization, and affect: Response to literary stories. Poetics 22: 389–407. Miall, D.S. & Kuiken, D. 1999. What is literariness? Three components of literary reading. Discourse Processes 28: 121–138. Miall, D.S. & Kuiken, D. 2001. Shifting perspectives: Readers‘ feelings and literary response. In New Perspectives on Narrative Perspective, W. van Peer & S. Chatman (eds), 289–301. Albany NY: State University of New York Press. Miall, D.S. & Kuiken, D. 2002. A feeling for fiction: Becoming what we behold. Poetics 30: 221–241. Nell, V. 1988. Lost in a Book. New Haven CT: Yale University Press. Oatley, K. 2002. Emotions and the story worlds of fiction. Narrative Impact: Social and Cognitive Foundations, M.C. Green, J.J. Strange & T.C. Brock (eds), 39–69. Mahwah NJ: Lawrence Erlbaum. Van Peer, W. 1986. Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm. Van Peer, W. 1997. Towards a poetics of emotion. In Emotion and the Arts, M. Hjort & S. Laver (eds), 215–225. Oxford: OUP. Van Peer, W. & Pander Maat, H. 1996. Perspectivation and sympathy: Effects of narrative point of view. In Empirical Approaches to Literature and Aesthetics, R.J. Kreuz & M.S. MacNealy (eds), 3–22. Norwood NJ: Ablex. Zwaan, R.A., Magliano, J.P. & Graesser, A.C. 1995. Dimensions of situation model construct ion in narrative comprehension. Journal of Experimental Psychology: Learning, Memory, Cognition 21: 386–197.
chapter 7
Two levels of foregrounding in literary narratives Yeshayahu Shen Discourse analysts and linguists have used the term “foregrounding” to refer to “… new information, in contrast to elements in the sentence which form the background against which the new elements are to be understood …” (van Peer & Hakemulder (2005). These foregrounded / backgrounded elements are linguistically marked by various “foregrounding / backgrounding markers”, used conventionally in standard language use. A few literary examples are analyzed which make deviatory use of those “foregrounding / backgrounding markers” (e.g., by marking central events in the story by “backgrounding markers”), yielding a “higher order foregrounding effect”, which corresponds to the notion of foregrounding as used by literary theorists. In this article, the relation between these two notions and levels of “foregrounding” is discussed. Keywords: foregrounding, foreground-background, narrative, literary discourse, deviation
1. The two notions of “foregrounding” “Foregrounding” has been, and still is, a central notion both in literary theory and in related disciplines, as testified by the most recent volume of Language and Literature edited by Willie van Peer (van Peer 2007). Not surprisingly, different disciplines have used this notion to mean different things, as rightly observed by van Peer & Hakemulder (2005). Two of the main uses have been developed in ‘linguistic/discourse analysis’ theories on the one hand, and ‘literary’ theories on the other. For linguists and discourse analysts, the term typically refers to “… new information, in contrast to elements in the sentence which form the background against which the new elements are to be understood by the listener / reader” (van Peer & Hakemulder 2005). These theories, known also as information staging or background-foreground articulation theories (e.g., Hopper & Thompson 1980; Reinhart 1984; for a survey see Unger 2002), have focused on standard uses
104 Yeshayahu Shen
of language in typically non literary discourse, and have pointed out an array of linguistic “foregrounding/backgrounding markers”, that is, a set of linguistic devices (at the syntactic, morphological, and lexical levels) whose function is to mark the information conveyed in the sentence as foreground or background information (I will return to this point below). Literary theories (e.g., Mukarovsky 1964; van Peer 2007), focusing on literary texts, used the notion of “foregrounding” in a broader sense. Those theories generally assume that outside literature there are automatic rules, maxims or conventions that have become automatized in language use. Literature aims at deviating from those automatized norms or conventions for creating various aesthetic effects, generally called ‘foregrounding effects’ (see van Peer 1986, 2007a). In general, these two different notions have been developed independently of one another, without any systematic attempt to relate the two. Note, however, that there is, at least, one potential interface between them that has not been given much attention: given that the standard use of “foregrounding markers”, typically in nonliterary discourse, is constrained by norms or conventions (described by linguists and discourse analysts), the latter could be thwarted or deviated from in literary discourse to create a foregrounding effect in the literary sense of the term, which would “enhance the meaning potential of the text, while also providing the reader with the possibility of aesthetic experience” (van Peer & Hakemulder 2005: p. 547). A case in point is the ending of Chekhov’s story “Sleepy” (1906/1971). The highlight event in this story is the baby’s murder by Varka, his teenage nanny. The story tells its life protagonist’s entanglements, and focuses on a specific night in her life. The night when the baby’s unstoppable crying interrupts her after work nap, embodies a miniature of her misery, and leads her to the kill and the following sleep. However, in absolute contrast to the event centrality, it is described in a linguistic form that marks it as being background or marginal event in the story: [1] Laughing and winking and shaking her fingers at the green patch, Varka steals up to the cradle and bends over the baby. When she has strangled him, she quickly lies down on the floor, laughswith delight that she can sleep, and in a minute is sleeping as sound as the dead
What is particularly noticeable here is that the most central event of the story – the strangling of the baby – appears as a subordinated clause (“When she . The English translation here, and in the subsequent examples maintains the syntactic properties of the original manuscript, relevant to the present analysis.
Chapter 7. Two levels of foregrounding in literary narratives
has strangled him …”), while the main clause of the sentence, refers to the marginal or background event of Varka’s lying down on the floor after the murder is completed. Clearly, this is a deviation from the standard use of the main / subordinate clause as described by various linguists and discourse analysts (e.g., Talmy 1978, Reinhart 1984; I will return to this point below), according to which the information in the foreground is supposed to appear in the main clause, while the information in the background should appear in the subordinate clause. However, precisely this deviation creates, at a higher level, a “foregrounding effect” in the sense used by literary theories, namely, deviation from a conventional norm or standard, for the creation of aesthetic effects. The goal of this paper is to illustrate this phenomenon, in which deviations from “lower level foregrounding norms” are created to achieve “higher level foregrounding effects”. In what follows I will introduce several examples, excerpted from various literary narratives that illustrate the deviations from the aforementioned norms, and the ways these deviations contribute to the overall meaning of the texts they were taken from.
2. Literary examples of deviations from foreground/background norms The first examples ([2] – [4]) are cases in which the deviations pertain to the norms regarding the use of the complex sentence. As already pointed out, the syntactic structure of complex sentences can, in some cases, mark the foreground/background distinction (Talmy 1975; Reinhart 1984; Hopper & Thompson 1980; Khalil 2001): the information coded in the subordinate clause is marginal compared to the information in the main clause.
. Consider, for example, the “lie test” proposed by Shir & Lappin (1979), that provides some evidence for this claim. The test puts the entire complex sentence in a direct speech context, and negates first the main clause, and then the subordinate clause. If it is not possible to negate the subordinate clause, it means that the main clause syntactic structure does not allow dominant interpretation of the subordinate. The following exchange illustrates the point: [1]: Y osi said: The man who promised to do well for the people, causes an inflation of 120% a year. 1a) L ie! He decreases the inflation. 1b) *Lie! He didn’t promise. The fact that 1b is a relatively strange response, indicates a syntactic structure that cannot be interpreted as dominant.
106 Yeshayahu Shen
The first case, which illustrates a point similar to the aforementioned Chekhov’s example is taken from the Israeli poet Avraham Shlonsky´s poem “Neum ploni al shkunato” (“One’s speech about his neighborhood” 1971). The poem describes different aspects of the poet’s city, characterized by an atmosphere of alienation and by hardly any mention of human existence emphasized by the detailed description of the number of several non-animate objects such as busses and buildings. At a certain point in the poem the following line appears, which is the first mention of the poem’s protagonist: [2] The one who had jumped off the window across / was satisfied with just three.
The most salient aspect of this description is the discrepancy between the content of the description, namely, the protagonist’s suicide, which is the most significant and shocking piece of information in the poem, and the insertion of this content into the syntactic position of a subordinate clause (“the one who …”).
3. S habtai and Be’er: Complex sentence and plotline representation versus divertive events Events that are part of the main plotline are more central than divertive events. Therefore, if these two kinds of events appear in the form of a complex sentence, the first would typically appear in the main clause, and the latter in the subordinate clause. A manipulative use of that norm is found in this next segment from “Past Continuous” (Shabtai 1977). Israel, who is in the room with Caesar remembers Ella and their past meeting: [3] They had met by chance at a movie, and although in the jeans and black sweater she wore underneath her dark green woolen coat she was no different from any other girl, she aroused in him from the first a feeling that she came from some unknown … place … as if she had materialized out of thin air and would disappear into it again one day … but he didn’t tell anyone about this feeling – certainly not [to] Caesar, who went into the second room (which served him as an office …). (9)
The sentence “to Caesar, who went into the second room” creates a (pragmatic) unacceptability. This effect is created by a contradiction between the backgroundforeground relations of the events, and the syntactic structure representing them: the event that is part of the main plotline (the entrance of Caesar which takes place
Chapter 7. Two levels of foregrounding in literary narratives
at the narrative present) is syntactically subordinated to a divertive or background scene which belongs to the narrative past. A similar phenomenon is exhibited by the following passage taken from “Feathers” (Be’er 1979): [4] Only once had I ever dared look at the memorial book for the victims of these riots, which lay hidden, to keep me from being traumatized by the sight of it, in a corner of our storeroom behind a photograph of my father’s first wife. Yet the face of the one-year-old Shlomo Slonim, who alone of all his family survived the butchery, still burned in my memory. It resembled the face of my father, who sat forsakenly now by the window, his head and shoulders striped with strange, delicate shadows … (17).
In this example there is a deviation from the sequence of events and a return to it. This return appears in a syntactic structure marking the relatively marginal information.
4. Amos Oz: The case of the conjunction marker “and” Another type of linguistic foregrounding marker is discourse connectives. One of the typical pragmatic functions of some of those connectives is marking (including linguistic morphemes that function as connectives) the relative importance of the discourse units they connect. A case in point is the conjunction marker “and”. A typical function of the connective “and” is to signal some equivalence in the level of “foregrounding” between the discourse units it connects (although see Ariel, (n.d.) The oddness of the following example is, arguably, the result of the violation of this conventional use of “and”. [5] At one of the allies he picked up a prostitute and followed her to a motel and stayed there until dawn … and read the festive newspaper … and pondered his thoughts and waited for the vicious dark to come and went out to the garden and hanged himself on a tree. (Oz 1974: p. 96)
The passage is characterized by a monotone reporting style that describes a series of routine, allegedly equally important actions. The “and” connective enhances this equivalence impression, but this stands in contradiction to the centrality of the most significant event of the story – that of the suicide of its protagonist.
108 Yeshayahu Shen
5. A mos Oz: The case of the definite article as a new/old information marker The new / old information distinction has been a major topic of research in linguistics and discourse analysis (see Prince 1979; Reinhart 1979; Ariel 1990). Various theories (e.g., Danesh 1974 & Firbas 1975) suggest that the new information constitutes the dynamic factor of communication, which advances the text, while the old information establishes new segments with previous text segments or with the communication situation. The new information, then, is in the foreground of each sentence in the discourse, while the old information represents the background against which the new elements are to be understood by the recipient of the sentence. Language has at its disposal an array of linguistic means by which this distinction is marked, such as pronouns, anaphors, and definite articles, which are typically used to mark a piece of information as old (Ariel 1990). The next two examples are taken from “Nomad and Viper” (Oz 1974), and may illustrate a violation of the old/new information marking norm. Two of the most central events in the story are related to two major encounters – the protagonist´s (Geula) meeting with the Bedouin nomad, and her encounter with the viper that causes her death. The centrality of these two events is directly suggested by the title of the story – Nomad and Viper. Consider, however, the first mention of the two encounters, described as follows: [6] The nomad stopped behind Geula’s back, silent as a fume.
[7]
Their flickers don’t intimidate the viper bustling near the girl’s body.
The two noun phrases referring to the two central referents (“the nomad” and “the viper”) are marked by the definite article “the” as old information, namely, as referents that have been previously introduced in the story. However, both referents are new referents that have not been mentioned prior to this point in the story, and what is even more crucial – they are central referents in that their encounters with the protagonist, Geula, are two of the most central events in the story plot. So, in both these cases, referents that have not been mentioned before and are central to the story plot are marked as old, that is, as background information. 6. The functionality of the deviations from foreground/background norms The deviations I have described are particularly significant for a literary theory of foregrounding as most of them tend to appear at central points of the story-plot,
Chapter 7. Two levels of foregrounding in literary narratives
such as the death of the protagonist, the new and surprising or shocking introduction of the protagonist, and so forth. Recall that the literary notion of foregrounding assumes that foregrounding “enhance the meaning potential of the text, while also providing the reader with the possibility of aesthetic experience” (van Peer & Hakemulder 2005). Clearly, the deviations described do achieve poetic effects of surprise and expectation violation, which is the case with all of the examples discussed above. However, in addition to these general effects, those deviations enhance the meaning potential of the text in various ways. Let us briefly examine three such examples. Recall that in Chekhov’s “Sleepy”, the information that is central to the whole story – Varka’s strangling of the baby – appears in a syntactically backgrounded position: “Laughing and winking and shaking her fingers at the green patch, Varka steals up to the cradle and bends over the baby. When she has strangled him, she quickly lies down on the floor, laughs with delight …”
This unusual way of introducing the protagonist’s highly extreme action can be viewed as corresponding with the narrator’s attempt to adhere to Varka’s point of view. This attempt is manifested throughout the story by the selection of information that is restricted to Varka’s point of view, the detailed description of Varka’s abuse by her employers as she experiences it, and, in particular, her extreme deprivation of sleep, eventually resulting in her hallucinatory state of mind. Varka’s physical and psychological exhaustion reaches its culmination at the point of the strangling of the baby. The syntax of the sentence in question (“when she has strangled him …”), serves the narrator’s attempt to enhance the reader’s empathy and identification with Varka given her horrifying, morally unacceptable action. The presentation of the baby’s strangling in the subordinate clause, and that of Varka’s immediate falling into sleep in the main clause, can be seen, hence, as the narrator’s device to produce the reader’s empathy towards Varka at this pivotal point of the story. In Shlonsky’s poem (1971) the deviatory use of the complex sentence consists of the use of the insertion of the thematically charged and dramatic event of the anonymous suicide into the subordinate clause. This results in the protagonist’s suicide as a “natural” or “necessary” result of the alienation, depression and boredom in his neighborhood. On the one hand, the grey routine, dominating the neighborhood, constitutes a suitable background to the created surprise upon encountering the suicide element. On the other hand, a closer look at the poem reveals that this is probably the main cause of the anonymous character’s suicide. The subordinate syntactic position presents the information concerning the suicide as old
110 Yeshayahu Shen
information in the poem context – and thus implies that it was concluded from the previous given information. In a way that suicide is to some extent a “necessary” result of the city’s characteristics. This impression is further established by the syntactic structure of the critical line, which presents it as background information. The other cases I discussed were taken from Amos Oz’s story – “Nomad and Viper”. Recall that in those cases the deviation from the norm was exhibited in the use of the definite article “the”, typically used to mark old (that is, background) referents, to represent newly introduced central referents. This deviatory use can be interpreted as reflecting the protagonist’s (Geula) wandering state of mind at the point of the two significant encounters with the nomad and the viper, taking place during her solitary walks outside her Kibuts’ territory. That is, in Geula’s unstable state of mind, the two wild and threatening invaders are part of her (literal and metaphorical) journey into no man’s land, stretching between the Kibuts and the Bedouin’s borderless living areas (see, e.g., Gertz 1980). The introduction of the nomad and the viper as old and background referents by the use of the definite article contributes to the representation of the protagonists’ inner world. To conclude, the examples analyzed in this article show that the relation between the two notions of foregrounding – the ‘linguistic’ and the ‘literary’, can directly contribute to both areas of research: the very oddity or pragmatic unacceptability of the above examples attests to the existence of the norms pertaining to ‘standard’ use of language as described by linguistic and discourse theories; from the standpoint of literary theories, however, the norms described by linguists and discourse analysts, serve as analytical tools necessary for the description of the intricacies and complexities of those literary phenomena.
References Ariel, M. 1990. The Function of Accessiblitity in a Theory of Grammar. London: Routledge. Ariel, M. (n.d.) And-associated inferences. Ms, Tel Aviv University. Be‘er, H. 1979. Feathers (Notzot). Tel Aviv: Am Oved. Chekhov, A. 1971. Selected Works for High Schools (Mivhar sipurey Chekhov le-vatey ha-sefer ha-tichoniim). Z. Zmiri (transl.). Tel Aviv: N. Tabersky. Danesh, F. 1974. Functional sentence perspective and the organization of the text. In Papers on Functional Sentence Perspective, F. Danesh F. (ed.), 107–128. Prague: Academia. Firbas, J. 1975. On the thematic and non-thematic section of the sentence. Style and Context: 317–334. Gertz, N. 1980. Amos Oz – Monography (Amos Oz – monografia). Tel Aviv: Sifriyat Poalim. Hopper, P. & Thompson, S. 1980. Transitivity. Language 56(2): 251–299. Khalil, E. 2001. Grounding and its signalling: Evidence from short news texts. Discourse Studies 3(1): 97–118.
Chapter 7. Two levels of foregrounding in literary narratives
Mukarovsky, J. 1964/1948. Standard language and poetic language. In A Prague School Reader, P.L. Garvin (ed.), 17–30. Washington DC: Georgetown University Press. Oz, A. 1974. Where the Jackals Howl. (Artzot hatan). 3rd ed. Tel Aviv: Massada. Reinhart, T. 1984. Principles of gestalt perception in the temporal organization of narrative texts. Linguistics 22(6): 779–809. Shabtai, Y. 1977/1994. Past Continuous. (Zikhron Devarim). Tel Aviv: Siman Kriah. Shlonsky, A. 1971. A. Shlonsky‘s Writings. (Kitvey A. Shlonsky). Tel Aviv: Sifriyat Poalim. Shen, Y. 1981. The hierarchy of information units in narrative texts. (ha-hirarchia shel yechidot informacia be-textim narrativim). MA Dissertation, Tel Aviv University. Shir, N. & Lappin, S. 1975. Dominance. Theoretical Linguistics 6: 43–88. Talmy, L. 1978. Figure and ground in complex sentences. In Universals in Human Language, J. Greenberg (ed.), 625–649. Stanford CA: Stanford University Press. Van Peer, W. 1986. Stylistics and Psychology. Investigations of Foregrounding. London: Croom Helm. Van Peer, W. (ed.). 2007. Language and Literature 16(2). Van Peer, W. 2007a. Introduction to foregrounding: A state of the art. Language and Literature 16(2): 99–104. Van Peer, W. & Hakemulder, J. 2005. Foregrounding. In Encyclopaedia of Language and Linguistics, Vol. 4, K. Brown (ed.), 546–551. Oxford: Elsevier. Unger, C. 2002. Global coherence, narrative structure, and expectations of relevance. Paper for the conference ‘Relevance Theory and Literature,’ University of Huddersfield, UK, Sept. 12–13.
chapter 8
Narrative empathy and inter-group relations János László & Ildikó Somogyvári In this study we assumed that the relation between readers’ and characters’ group identity would influence narrative empathy and thereby the impact of a short story. We used three social psychological models (infrahumanization, mentalization and linguistic inter-group bias) to test our assumptions. Methodologically, we used a narrative recall paradigm based on the assumption that narrative recall carries also the experiential aspects of the text processing and thereby enables a fine grained analysis of meaning construction. We also measured liking of the story as an impact variable and the strength of national identification as a moderator variable. Results only partly supported our assumptions. Whereas empathy and liking are strongly correlated, Hungarian subjects overall did not feel more empathy with the characters of the Hungarian story version, did not like more this story, did not assign more secondary emotions to the Hungarian characters, and did not describe positive behavior of the Hungarian characters more abstractly then happened in the case of the “Slovak story” version. Keywords: narrative empathy, narrative perspective, infrahumanization, linguistic inter-group bias, identity
1. Narrative perspective The content of a narrative, including elements such as events, characters and circumstances, must be presented from a point of view (Prince 1987). Narrative perspective can be thought of as a relational concept between the producer and recipient of narrative (Bal 1985). According to Wiebe (1991), this is communicated by the distance in time and space the author takes vis-à-vis the content, and by the possibility that
. This study was supported by the grants OTKA 49413/2006 and NKFP6–00074/2005 to the first author.
János László & Ildikó Somogyvári
the narrator may express a character’s beliefs, emotions or evaluations. This latter component of a perspective is sometimes called a psychological perspective (e.g., Uspensky 1974) or internal perspective (Booth 1983). Through these components, narrative perspective establishes a surface structure empathy hierarchy (Kuno 1976) that influences how the reader or listener constructs the meaning of the narrated event and it opens the way for participatory affective responses (Gerrig 1996). The role of perspective taking is often demonstrated in cognitive studies, particularly in memory tasks (e.g., Anderson & Pichert 1978; Owens, Bower & Black 1979), but its effects are pointed out in relation to other cognitive phenomena. For instance, the actor-observer attributional difference known from the field of social cognition (Jones & Nisbett 1972) can be achieved by manipulating the perspective given in the instructions. Subjects reading the same story will identify a different character as the main character depending on the instructions given. Regardless of which character they thought of as the main character or actor, subjects reading the story will explain the action of the chosen character with situational rather than dispositional factors. This phenomenon is explained by? cognitive empathy initiated by perspective taking. Studies with literary narratives, however, do not always support the impact of the narrative perspective. Whereas Andringa (1986) and László (1986) found strong effects, other authors, e.g., Ludwig & Faulstich (1985) or Schramm (1985) found week effects or no effects at all (see van Peer & Pander Maat 1996 for a review). In a study with Dutch teenagers van Peer & Pander Maat (1996: 152) found that first person narration created a greater illusion of closeness and allowed the readers a “greater and better fusion with the world of the character”, but they also showed that the enhancement of sympathy for protagonists through positive internal focalization actually weakened as teenagers matured. In a study, László & Larsen (1991) examined the qualities of the phenomenal experience of subjects when reading literary texts in a Danish-Hungarian comparative study. Internal-external perspective was varied in the texts. By the help of a non-intrusive quasi-online methodology, the so-called self-probed retrospection (Larsen & László 1990) subjects had to recollect memories, which occurred during reading. They also had to rate the phenomenal qualities of these qualities such as vividness or aggressiveness. Memories recollected by Danish and Hungarian subjects during reading did not significantly differ in number. On the other hand, content and characteristics of the recalled experience(s) did differ. This difference, however, could be ascribed exclusively to the cultural variable, because the perspective variable had no influence on the psychological content of the recalled material. The only dimensions where focalization exerted impact independently from the cultural effects were phenomenal qualities of the memories. Both aggressiveness and vividness of the memories were affected a great deal by perspective: segments with internal perspective resulted in recollection of more aggressive and more vivid memories.
Chapter 8. Narrative empathy and inter-group relations
Our results, just like those of van Peer & Pander Maat (1996) suggest that cognitive and emotional empathy with characters of a literary narrative are governed not only by the narrative perspective used by the author of the text. There are factors such as cultural background of the reader, or stage of the identity development, which may influence allocation of empathy and thereby the way readers construct the meaning of the text.
2. Narrative empathy Psychological models of empathy (Davis 1994; Eisenberg & Strayer 1987; Preston & de Waal 2002) broadly define the concept of empathy as any process of attending to another’s state in a way that induces a state in oneself more appropriate to the situation of the object attended to than to one’s own situation. Empathy has cognitive and emotional components. Cognitive empathy embraces a wide array of responses from elementary circular reactions (Piaget 1983) through role taking (Mead 1934) to complex cognitive processes such as elaborated cognitive networks (Eisenberg & Stayer 1987). Emotional empathy includes parallel emotional processes, when the emotional state of the observer corresponds to the state of the observed person and reactive emotional processes, when the observer experiences an emotion which is a consequence of the emotion of the observed person. This latter type of emotional empathy is often called sympathy (e.g., Keen 2006; van Peer & Pander Maat 1996). Empathy with narrative characters can be facilitated or inhibited not only by narrative devices such as internal or external perspectives. Gerrig (1996) points out that readers are inclined to categorize characters, and assign category-based dispositions to them, whereas Hogan (2001) speaks about categorical empathy (as opposed to situational empathy), when readers’ identity matches that of the characters’ and empathic processes are facilitated by this categorical identity. Whereas situational empathy always requires a reader’s having a memory of a comparable experience, categorical empathy is based on generalized group experiences and may result in short cuts in interpretation and subsequent emotional reactions. Susanne Keen’s theory of narrative empathy (Keen 2006) traces back the differences between readers’ responses to different work of empathy in the text-reader interaction. She claims that the position of the reader with respect to the author’s strategic empathizing in fictional world making limits the reader’s empathy. She distinguishes between three types of narrative empathy. Bounded strategic empathy operates within an in-group, stemming from experiences of mutuality and leading to feeling with familiar others. Ambassador strategic empathy addresses chosen others with the aim of cultivating their empathy for the in-group, often to a specific end. Broadcast strategic empathy calls upon every reader to feel with the
János László & Ildikó Somogyvári
members of a group, by emphasizing common vulnerabilities and hopes through universalizing representations. It follows from this theory that narrative empathy, particularly bounded strategic empathy may serve the goals of the reader’s own group by strengthening group identity and facilitating prosocial behavior towards group members. Narratives of historiography and historical novels are particularly prone to introduce bounded strategic empathy. In a study in which we compared sections of Austrian and Hungarian history textbooks dealing with the Austro-Hungarian Monarchy, we showed that Hungarian textbooks, by using a wide range of rhetoric devices, e.g., mentalization of the actions of the Hungarian characters, exploit this opportunity more strongly than Austrian textbooks do (Vincze, Tóth & László 2007). In other studies we showed that authors of Hungarian historical novels systematically use devices of social psychological composition. In-group characters are depicted along a single salient positive attribute, which makes empathy and identification with the characters readily available. The fact that positive values are distributed among in-group members facilitates identification with the group (László & Vincze 2004). László, Vincze & Kőváriné Somogyváry (2003) found that not only in-group – out-group evaluations differed in Hungarian historical novels, but also positive versus negative outcomes of the actions of the in-group and that of the out-groups. Despite the fall of a revolution and a freedom fight, in-group members mostly succeeded in their actions whereas the adversaries mostly failed. Based on these results, Liu and László (2007) suggested that studying narrative empathy is a royal way to explore relations between group identity and historical representations. Keen (2006) also urged to study emphatic reactions with regard to group identity. Very few empirical explorations have been conducted in this field. The study we present below focuses on narrative empathy, both cognitive and emotional, by varying the group identity of the main characters. Three social psychological models of group identification will be tested through indirect measures of group perception. First we introduce the three models briefly. a. Infrahumanization The tendency for people to favor their in-group in comparison to an out-group is well documented in experimental social psychology (see Brewer & Brown 1998 for a review). Leyens et al. (2000) argue that if people are biased toward favoring their in-group, then people should choose the best essences (i.e., human essences) for their own group. Leyens and his colleagues refer to this as infrahumanization, the tendency for people to perceive their own group as more human in comparison to out-groups. One of the symptoms of infrahumanization is that group members reserve the uniquely human so called social or secondary emotions for their own group, whereas out-groups are endowed with primary emotions that humans
Chapter 8. Narrative empathy and inter-group relations
share with animals. This phenomenon has been demonstrated across a number of studies in several cultures using various methodologies (Demoulin, Leyens, Paladino, Rodriguez, Rodrigues & Dovidio 2004; Leyens et al. 2001; Paladino, Leyens, Rodrigues, Rodrigues, Gaunt & Demoulin 2002). b. Mentalization According to recent results, infrahumanization involves not only emotions, but also the use of human-related versus animal-related concepts (Viki, Winchester, Titshall, Chisango, Pina, & Russel 2006), and mental concepts. Kozak, Correll, & Doan (2008) showed that subjects who scored high on a prejudice scale attributed significantly less mental states to colored target persons than to members of their own social category. These results are also related to the problem of perceived similarity. Attributing mental states to others always involves perspective taking. In-group members are usually perceived as more similar to the self than? out-group members are, therefore it is easier to take their perspective. Dissimilar out-group members’ behavior is explained in terms of stereotypes rather than of mental states accessed by perspective taking (Ames 2004; Galinsky & Moskowitz 2001). c. Linguistic inter-group bias (LIB) Linguistic inter-group bias is based on the Linguistic Category Model (Semin & Fiedler 1988, 1991). This model suggests that the abstraction level of linguistic categories is related to causal attribution of behavior. The model singles out four abstraction levels: descriptive action verbs (e.g., hits), interpretative action verbs (e.g., hurts), state verbs (e.g., misunderstands) and adjectives (e.g., aggressive). Using low abstraction level (e.g., John hits Paul or John hurts Paul) suggests situational attribution, whereas high abstraction level (e.g., John misunderstands Paul or John is aggressive) suggests dispositional attribution. The model has recently been amended with a fifth abstraction level of State action verbs (e.g., amazes). Linguistic inter-group bias occurs when in-group members describe their own positive behavior abstractly, whereas they tend to describe similar behavior of the out-group in concrete terms. The former enables dispositional attribution, whereas the latter elicits situational attribution, i.e., prevents generalization across situations.
3. The experimental study 3.1 Material A twentieth century Hungarian short story, “The Day of the Battle” by Zsigmond Móricz (a well known Hungarian author) was selected for the study. The story is written in a realist style. The main characters are an old farmer and his wife.
János László & Ildikó Somogyvári
Two soldiers trust them to take care of a flock of sheep before a battle begins in the neighborhood. The soldiers die in the battle and the flock stays with the old people. While dealing with the sheep, the old man gets hit by a bullet by accident and dies. The story does not lay out the inter-group conflict (actually the HungarianAustrian conflict in the 1848–49 revolution and freedom fight), but focuses instead on the inner life of the characters. The story was transcribed into two versions. In the “In-group (Hungarian) version” only temporal anchors were deleted, so as to prevent readers from contextualizing temporally the story. For the “Out-group” version we choose a group with which Hungarians have a historical rivalry. In this version not only were temporal anchors\deleted, but also all the names and locations were transcribed into Slovak. 3.2 Procedure Forty eight Hungarian subjects participated in the study in a classroom situation. They were instructed that they would be participating in a Central-European research project on historical novels and short stories. Half of the subjects read the “Hungarian” version of the short story, the other half read the “Slovak” version. After reading they filled out a questionnaire consisting of 7 or 5 points scales. Scales included liking (5 point scale), empathy with the characters (7 point scale), and aesthetic value (5 point scale). In the next step, they were asked to recall the text of the short story in writing. Finally, they filled out the Hungarian version of the Doosje, Branscombe, Spears & Manstead (1998) questionnaire on identi fication with nation, which included questions such as “I am proud of being Hungarian” or “I miss my country if I am abroad”. 3.3 Data analysis and results We calculated aggregate measures of the 6 empathy scales and 10 scales of national identification. These results as well as liking scores are shown in Table 1. Table 1. Average scores of empathy, national identification and liking scales
Narrative empathy
Mean “Hungarian” version “Slovak” version Total
25.08 26.12 25.6
National identification
Std. deviation Mean 3.45 4.76 4.15
58.25 57.17 57.71
Liking
Std. deviation Mean 10.57 11.49 10.93
3.42 3.67 3.54
Std. deviation 0.77 0.87 0.82
Chapter 8. Narrative empathy and inter-group relations
There were no significant differences between the “Hungarian” and the “Slovak” versions of the story in either dimension. 3.3.1 Analysis of the recalls Length of the recalls varied between 232 and 475 words. In the case of the “Hungarian” story the total word count was 5,064 words and this number was almost the same with the “Slovak” story. There was no difference in the recalled events between the two stories. Readers of the two text versions remembered of the same events from the text with practically identical frequency. a. Emotions and mental states Two independent coders rated the emotions and the mental states in the recalls with a .83 agreement. We used the Hungarian emotion dictionary for coding (Fülöp & László 2006). This dictionary classifies emotions into four categories: primary emotions (e.g., happiness), secondary emotions (e.g., shame), affects (any changes in activation contour, e.g., calms down), and feelings (good or bad moods). The dictionary, with the help of local grammars, contains all the inflected versions of the emotion words. A similar mental state dictionary was used to code mentalization (Vincze & László 2006). Given that only secondary emotions are uniquely human, based on the infrahumanization theory we expected more secondary emotions in the “Hungarian story”-recalls, then in the “Slovak story”-recalls, whereas we did not expect differences in the other three types of emotions. Similarly, for mentalization, we expected more mental states attributed to the Hungarian characters than to the Slovak characters. Frequency data are presented in Table 2. Table 2. Frequency data of the emotions and mental states in the recalls
Frequency of primary emotions
Frequency of secondary emotions
Frequency of affects
Frequency of mental states
Std. Std. Std. Std. Mean deviation Mean deviation Mean deviation Mean deviation “Hungarian” version “Slovak” version Total
2.79
2.67
0.75
0.94
0.62
0.97
4.5
2.96
2.7 2.75
1.33 2.09
0.71 0.73
0.8 0.87
0.54 0.58
0.93 0.94
5.71 5.1
3.21 3.12
Frequency data show that although there are slightly more emotions, including secondary emotions, in the “Hungarian story” recalls, the difference is not significant
János László & Ildikó Somogyvári
with either emotion type (non-parametric Mann-Whitney tests). We received the same results when we adjusted the frequencies to the length of the recalls. b. Linguistic inter-group bias Linguistic categories in the recalls were coded by two independent judges with a .87 agreement. Verbs and adjectives were coded not only according to linguistic abstraction, but also according to the “moral” value of the action or attribute (positive, negative, and neutral). Only those linguistic expressions were coded which explicitly belonged to one of the characters. The coding was based on the Hungarian evaluative lcm dictionary (Bigazzi & Nencini 2006). Results are shown in Table 3. Table 3. Frequency of linguistic categories in the recalls
“Hungarian” Sample
“Slovak” Sample
Total
Std. Std. Std. Mean deviation Mean deviation Mean deviation Descriptive Action verbs DAV
Positive 0.37 Negative 1.41 Neutral 12.67
0.65 0.93 5.8
1 1.54 11.04
1.02 0.78 7.05
0.69 1.48 11.85
0.9 0.85 6.44
Interpretative Action verbs IAV
Positive Negative Neutral
0.79 1.37 1.54
0.88 1.31 1.18
1.08 1.17 1.17
1.02 1.27 1.13
0.94 1.27 1.35
0.95 1.28 1.15
State Action verbs SAV
Positive Negative Neutral
0.12 0.58 1.08
0.34 0.88 0.88
0.17 0.33 0.83
0.64 0.64 0.56
0.14 0.46 0.96
0.5 0.77 0.74
State verbs
Positive Negative
0.79 0.95
0.88 1.04
0.79 1.33
0.98 1.09
0.79 1.14
0.92 1.07
SV
Neutral
3
2.36
4.04
2.58
3.52
2.5
Adjectives ADJ
Positive Negative
0.21 1.04
0.58 1.23
0.12 0.58
0.34 0.93
0.17 0.81
0.47 1.1
1.73
0.22
1.83
0.27
1.78
0.25
Abstraction level
Based on the linguistic inter-group bias model, we expected higher level of linguistic abstraction for the positive deeds in the recalls of the “Hungarian story” (in-group version). A similar level of high linguistic abstraction was expected for the negative deeds in the recalls of the “Slovak story” (out-group version). The results did not confirm our expectations. There was not any significant difference between the recalls of the two groups either for positive or for negative behavior, except that readers of the Slovak story version gave more positive DAVs in their recalls (p < 0.01, Mann-Whitney non-parametric test).
Chapter 8. Narrative empathy and inter-group relations
c. Correlations with independent measures of empathy, national identification and liking Considering the contradictory results, we collapsed the two samples and calculated Pearson correlations to see whether the independent variables of narrative empathy, readers’ national identification and liking would show mediating effects to infrahumanization and linguistic abstraction measures. Results are shown in Table 4. Table 4. Independent measures: empathy. national identification and liking (total sample) Variables Narrative empathy x Liking Narrative empathy x Frequency of secondary emotions Narrative empathy x Relative frequency of secondary Emotions Narrative empathy x Relative frequency of mental states Frequency of primary emotions x Frequency of mental states Frequency of primary emotions x Frequency of affects
Pearson correlation
Sig (2–tailed)
0.518
< 0.01**
0.347
0.016*
0.32
0.027*
-0.315
0.029*
0.37
0.01*
0.335
0.02*
*Correlation is significant at the 0.05 level. **Correlation is significant at the 0.01 level3.
It is not surprising that measures of narrative empathy and liking correlate with each other. More interesting is the fact (result?) that narrative empathy also showed a significant correlation with frequency of recalled secondary emotions and with relative frequency of recalled mental states. The correlations between primary emotions and recalled mental states as well as the correlations between primary emotions and affects are again plausible, particularly the latter one, because the two emotion types do not differ in the humanness dimension. National identification did not show any significant correlation with either of the other variables. Next we performed correlations in the two samples separately. Results with the “Hungarian story” are shown in Table 5. Beyond the plausible correlations between narrative empathy and liking, and between the frequency of recalled primary emotions and affects, which correlations were also observed with the total sample, the “Hungarian story” sample showed two somewhat contradictory correlations. On the one hand, narrative empathy correlated with the relative frequency of recalled secondary emotions, mental states and affects. This is in accord with the results of the total sample. On the other hand, a strong correlation occurred between national identification and liking. Similar correlation analysis was done with the “Slovak story” sample. Results are shown in Table 6.
János László & Ildikó Somogyvári
Table 5. Independent measures of empathy, national identification and liking (“Hungarianstory” sample) Variables Narrative empathy x Liking National identification x Liking Narrative empathy x Frequency of secondary emotions Narrative empathy × Relative frequency of mental states Narrative empathy × Relative frequency of affects Frequency of primary emotions × Frequency of mental states Frequency of primary emotions × Frequency of affects
Pearson correlation
Sig (2–tailed)
0.523 0.708
0.009** < 0.01**
0.394
0.057 (tendency)
-0.43
0.036*
0.36
0.004**
0.486
0.016*
0.422
0.04*
*Correlation is significant at the 0.05 level. **Correlation is significant at the 0.01 level.
Table 6. Correlations with independent measures of empathy, national identification and liking in the “Slovak-story” sample Variables Narrative empathy × Liking Narrative empathy × Relative frequency of secondary emotions
Pearson correlation
Sig (2–tailed)
0.504
0.012*
0.401
0.052 (tendency)
*Correlation is significant at the 0.05 level.
The only significant correlations, at least on tendency level, could be observed between narrative empathy and liking and between narrative empathy and relative frequency of recalled secondary emotions. These results are in accord with the results of the total sample. However, no other correlations which occurred with the total sample or the “Hungarian story” sample were present.
4. Discussion In this study we assumed that the relation between readers’ and characters’ group identity would influence narrative empathy and thereby the impact of a short
Chapter 8. Narrative empathy and inter-group relations
story. We used three social psychological models (infrahumanization, mentalization and linguistic inter-group bias) to test our assumptions. Methodologically, we used a narrative recall paradigm based on the assumption that narrative recall carries also the experiential aspects of the text processing and thereby enables a fine grained analysis of meaning construction (see László 2008). We also measured liking of the story as an impact variable and the strength of national identification as a moderator variable. Results only partly supported our assumptions. Whereas empathy and liking strongly correlated, Hungarian subjects overall did not feel more empathy with the characters of the Hungarian story version, did not like this story more, did not assign more secondary emotions to the Hungarian characters, and did not describe positive behavior of the Hungarian characters more abstractly than in the case of the “Slovak story” version. The theory of infrahumanization implies that more empathy is directed to people to whom we ascribe more secondary emotions and mental states. Behavioral consequences of the infrahumanization can be observed in the empathy based prosocial behavior (Leyens et al. 2000). Our results suggest a reversed relation. Subjects who felt more empathy with the characters recalled more secondary emotions and mental states, i.e., “humanized” the characters more independently of the characters’ in-group or out-group affiliation. Even linguistic abstraction was more related to empathy than to group identity. These results suggest a rather intricate working of empathy, which can be interpreted in the model of Keen (2006). We assume that an equivalent of the author’s bounded strategic empathy was at work primarily in readers characterized with high national identification. These readers liked the “Hungarian story” most and their liking correlated with their empathy with the characters of their own group. The rest of the readers could be characterized with empathic processes, which correspond to the other two types of authorial strategic empathy: ambassador strategic empathy and broadcast strategic empathy. It means that these readers allocated their empathy not only to in-group members, and, by experiencing outgroup characters’ hopes and vulnerabilities as common, they universalized their representations. These readers can take off from the narrow in-group perspective and adopt the perspectives of the characters who belong to different out-groups. Thereby they can do “justice” to perspectives opened by authors’ narrative strategies (see van Peer 2001). The lack of most of the expected social psychological effects can by explained with a relatively low prevalence of empathic processes that correspond to bounded strategic empathy. Had we chosen a story in which a strong inter-group conflict is depicted, we would probably have found stronger effects of in-group empathy.
János László & Ildikó Somogyvári
References Anderson, R.C. & Pichert, J.W. 1978. Recall of previously unrecallable information following a shift in perspective. Journal of Verbal Learning and Verbal Behavior 17: 1–12. Andringa, E. 1986. Perspektivierung und Perspektienübernahme. spiel 5: 135–146. Booth, W.C. 1983. The Rhetoric of Fiction. 2nd edn. Chicago IL: Chicago University Press. Brewer, M.B. & Brown, R.J. 1998. Intergroup relations. In Handbook of Social Psychology, 4th edn, Vol. 4, D.T. Gilbert, S.T. Fiske & G. Lindzey (eds), 554–594. New York NY: McGrawHill, Davis, M.H. 1994. Empathy. Madison WI: Brown and Benchmark. Demoulin, S., Leyens, J.-P., Paladino, M.P., Rodríguez, T.R., Rodríguez, P.A. & Dovidio, J.F. 2004. Dimensions of uniquely and non-uniquely human emotions. Cognition and Emotion 18(1): 71–96. Doosje, B., Branscombe, N.R., Spears, R. & Manstead, A.S.R. 1998. Guilty by association: When one’s group has a negative history. Journal of Personality and Social Psychology 75: 872–886. Eisenberg, N. & Strayer, J. (eds) 1987. Empathy and its Development. Cambridge: CUP. Fülöp, É. & László, J. 2006. Az elbeszélések érzelmi aspektusának vizsgálata tartalomelemző program segítségével. (Studying emotions in stories through computer algorithms) IV. Magyar Számítógépes Nyelvészeti Konferencia konferenciakötete, Juhász Nyomda, Szeged, 296–304. Galinsky, A.D. & Moskowitz, G.B. 2000. Perspective taking: Decreasing stereotype expression, stereotype accessibility, and in-group favoritism. Journal of Personality and Social Psychology 78: 708–724. Gerrig, R.J. 1993. Experiencing Narrative Worlds. New Haven CT: Yale University Press. Jones, E.E. & Nisbett, R.E. 1972. The actor and the observer: Divergent perceotions of the causes of behavior. In Attribution: Perceiving the Causes of Behavior, E.E. Jones, D.E. Kanouse, H.H. Kelley, R.E. Nisbett, F. Valins & B. Weiner (eds), 79–94. Morristown NJ: General Learning Press. Keen, S. 2006. A theory of narrative empathy. Narrative 14(3): 207–236. Kozak, M.N., Correll, J. & Doan, T. (In preparation). Mind attribution and prejudice. Kuno, S. 1976. Subject, theme and the speaker’s empathy – A reexamination of the relativization phenomena. In Subject and Topic, C.N. Li (ed.), 47–51. New York NY: Academic Press. Larsen, S.F. & László, J. 1990. Cultural-historical knowledge and personal experience in appreciation of literature. European Journal of Social Psychology 20: 435–440. László, J. 1986. Same story with different point of view. SPIEL 5: 1–22. László, J. 2008. The Science of Stories. Introduction into Narrative Psychology. London: Routledge. László, J. & Larsen, S.F. 1991. Cultural and text variables in processing personal experiences while reading literature. Empirical Studies of the Arts 9: 23–34. László, J. Vincze, O. & Kőváriné Somogyvári, I. 2003. Representation of national identity in successful historical novels. Empirical Studies of the Arts 21(1): 69–80. László, J. & Vincze, O. 2004. Coping with historical tasks. The role of historical novels in transmitting psychological patterns of national identity. spiel 21(1): 76–88. Leyens, J.P., Paladino, P.M., Rodriguez, R.T., Vaes, J., Demoulin, S., Rodriguez, A.P. et al. 2000. The emotional side of prejudice: The role of secondary emotions. Personality and Social Psychology Review 4: 186–197. Liu, J.H. & László, J. 2007. A narrative theory of history and identity: Social identity, social representations, society and the individual. In Social Representation and Identity; Content, Process and Power, G. Moloney, Walker, I. (eds) New York NY: Palgrave Macmillan.
Chapter 8. Narrative empathy and inter-group relations
Ludwig, H.W. & W. Faulstich 1985. Erzahlperspektive empirisch. Tübingen: Günter Narr. Maas, A., Salvi, D., Arcuri, L. & Semin, G.R. 1998. Language use in intergroup contexts: The linguistic intergroup bias. Journal of Personality and Social Psychology 57: 981–993. Mead, G.H. 1934. Mind, Self, and Society. Chicago IL: Chicago University Press. Owens, J., Bower, G.H. & Black, J. 1979. The “soap-opera” effect in story recall. Memory and Cognition 7: 185–191. Piaget, J. 1983. Piaget’s theory. In Handbook of Child Psychology. Vol. 1. History, Theory and Methods, P.H. Mussen (ed.), 247–286. New York NY: Wiley. Preston, S.D. & de Waal, F.B.M. 2002. Empathy: Its ultimate and proximate bases. Behavioral and Brain Sciences 25(1): 1–72. Prince, G. 1982. Narratology: The Nature and Function of Narratives. The Hague: Mouton de Gruyter. Schramm, D. 1985. Norm en normdoorbreking. Amsterdam: VU Uitgeverij. Semin, G.R. & Fiedler, K. 1988. The cognitive functions of linguistic categories in describing persons: Social cognition and language. Journal of Personality and Social Psychology 54: 558–568. Semin, G.R. & Fiedler, K. 1991. The linguistic category model, its bases, applications and range. In European Review of Social Psychology, Vol. 2, W. Stroebe & M. Hewstone (eds), 1–30. Wiley: Chichester. Van Peer, W. & Pander Maat, H. 1996. Perspectivation and sympathy. Effects of narrative point of view. In Empirical Approaches to Literature and Aesthetics, R.J. Kreutz & M.S. MacNealy (eds), 143–154. Norwood NJ: Ablex. Van Peer, W. 2001. Justice in perspective. In New Perspectives on Narrative Perspective, W. van Peer & S. Chatman (eds), 325–338. Albany NY: SUNY Press. Viki, T.G., Winchester, L., Titshall, L., Chisango, T., Pina, A. & Russel, R. 2006. Bexond secondary emotions: The infrahumanization of outgroups using human related and animal-related words. Social Cofnition 24(6): 753–775. Vincze, O. & László, J. 2006. IV. Magyar Számítógépes Nyelvészeti Konferencia konferenciakötete, Juhász Nyomda, Szeged, 296–304. Vincze, O., Tóth J., & László J., 2007. Representations of the Austro-Hungarian monarchy in the history books of the two nations. ETC – Empirical Text and Culture Research 3: 62–71. Uspensky, B.A. 1974. The Poetics of Composition: Structure of the Artistic Text and the Typology of Compositional Forms. Berkeley CA: University of California Press. Wiebe, J. 1994. Tracking point of view in narrative. Computational Linguistics 20(2): 233–287.
chapter 9
Effects of reading on knowledge, social abilities, and selfhood Theory and empirical studies Raymond Mar, Maja Djikic & Keith Oatley Reading exhibits a principle of expertise: the more one does it the more skilled one is likely to become both in the activity and in content knowledge. Our experiences with text lead to the acquisition of both vocabulary and general knowledge. Research from our group examines how reading can have other outcomes. With a starting point of fiction as an entryway into simulations of social interactions, we review empirical studies of how the reading of fiction can improve empathy and other social abilities, and prompt changes in personality. Keywords: fiction, reading, simulation, empathy, personality change, theory-of-mind
1. Introduction If science exemplifies the exploring mind of the academy, literature remains its heart. Few have done as much as Willie van Peer to maintain the heart, and at the same time to apply the mind to the study of literature. In his research on point of view and sympathy (van Peer & Maat 1996) and on foregrounding (van Peer 1986, 2007), he has shown how we can deepen our understanding of central aspects of literariness–the heart of literature–and offer evidence in the place of opinion. In this chapter, we follow van Peer’s example of empirical exploration and raise the question of whether and how reading can change the reader. We use the theory of expertise as a basis of thinking about how reading can have psychological effects that continue when one puts the book down. The main method employed to understand how skills are attained has been the study of expertise. This research in cognitive psychology (e.g., Ericsson 1990,
Raymond Mar, Maja Djikic & Keith Oatley
Ross 2006) has shown that to become an expert – a person of accomplishment – in any skill, one must devote at least 10,000 hours to the domain of interest. This works out to three hours a day for ten years (taking account of holidays and weekends, this is about the amount of time children spend in school classrooms during their lives) or ten hours a day for three years (about the time undergraduates spend completing a university degree). For maximum accomplishment, the time is best spent in problem solving, and in acquiring knowledge and procedures in a particular domain. Coaching is often an important component. The acquisition of the many skills of reading falls readily under this rubric (Wagner & Stanovich 1996). For perhaps two thousand years after the invention of writing, the activity of reading and writing was the province mainly of scribes who worked with administrators. Then, about 2500 years ago in Europe, the learning of reading and writing by wider sections of the population began with the invention of writing in an alphabetic language, Greek (Powell 2002). Since then, coaching in the skills of reading and writing has gradually become more widespread. Today it is the principal task of the world’s education systems. The general term to designate completion of a school education is the achievement of literacy: being able to read. Literacy has huge effects on society. It is a prerequisite for the many technologies – in industry, housing, power generation, transport, commerce, health, and information – on which advanced societies have become utterly dependent. But what about the effects of reading on the individual? These are by no means so well recognized. The most important research program on such effects has been conducted by Stanovich, West, and their colleagues (e.g., West, Stanovich & Mitchell 1993; Stanovich 1993; Stanovich, West & Harrison 1995; Echols et al. 1996). The program involved two steps. First, Stanovich, West, and their colleagues developed a method for assessing how much people read in their daily lives. To start with it seemed clear that to see how much people read, daily diaries of activities would need to be kept. But this was laborious. Stanovich & West (1989), therefore, invented a checklist of the names of authors of books: the Author Recognition Test (art). As well as names of authors the list included, as foils, names of people who were not authors. Participants were asked to check all those names they knew to be authors, and a score was derived by subtracting the number of foils from the number of real authors. People who read a lot know the names of authors from their own reading and from reviews, visiting bookshops, and so on. The art is easy to administer and score, and there are versions for adults and children. Scores on these tests were found to be very good proxies for diary measures of reading, and indeed to correlate well with behavioral observations of the amount of reading people did (Stanovich 1993). It has thus become the method of choice for determining the extent of people’s reading. The general term for the measure is “print exposure.”
Chapter 9. Effects of reading on knowledge, social abilities, and selfhood
Second, these researchers took as outcome measures people’s vocabulary, verbal skills, and the amount they knew in various domains of knowledge. Then, with these outcomes, they used the method of hierarchical regression and entered into the equation first such measures as age, social class, and general intelligence score (iq), and then measures of print exposure, such as the art. Even when age, social class, and iq were controlled for, print exposure was a strong predictor of vocabulary, language use, and general knowledge. The more you read, the more you know and the better you know it. Interestingly, print exposure does not predict everything in the cognitive domain. For instance, Siddiqui, West & Stanovich (1998) showed that although print exposure has been found to be a good predictor of word usage, it is not a good predictor of how to use words in a de-contextualized way to reason in syllogisms. Our research has, as it were, taken off from the psychology of expertise and from the methods and results of Stanovich, West, and their colleagues. We have sought to understand the effects not of reading in general, but of reading fiction. The acquisition of knowledge is a logical outcome of reading non-fiction, but what results from the reading of fictional literature? Is fiction just a pastime, an entertainment, or does it have psychological effects that can be distinguished from those of reading non-fiction? Does reading the works of great artists have effects that can be distinguished from reading the same information but without artistic form? The theory of fiction from which we start is that a novel, short story, play, or film is a kind of simulation that runs not on computers but on minds (Oatley 1999). The simulation is both of other minds, and of people’s interactions in the social world. We argue that people are good at understanding processes one step at a time, but are much less good at understanding interactions of these processes with others. Thus in thinking about the weather, we can understand that winds blow from areas of high atmospheric pressure to areas of low pressure. But what happens when other factors operate? Does the simple understanding hold when a warm mass of air is blown towards a cold mass? Does it operate in the same way when winds pass over land and over water? To help understand such interactions of multiple factors we need simulations. Hence to give a weather forecast we need to enter into a computer simulation both the wind-producing effects of different atmospheric pressures and also many other processes that interact with them. So, when you look at a map or summary of tomorrow’s weather on the television or in the newspaper, you are looking at the output of a computer simulation. Similarly, we argue, it is easy to understand single factors in the social world. We know that if someone, say Alice, is thwarted in a strong desire by Beatrice, Alice is likely to be angry with Beatrice. But what happens when Beatrice is Alice’s boss? What happens when Beatrice is Alice’s daughter? What happens when Beatrice is Alice’s
Raymond Mar, Maja Djikic & Keith Oatley
lover? Novels, short stories and plays enter basic tendencies such as instigations to anger into simulations with other social processes that can affect them. In the way that the psychology of expertise has shown that practice is important for any skill, we argue that to read or watch such simulations in books and theaters is to set ourselves social problems, and practice on them. We might say that just as a cockpit simulator sets problems and gives practice in piloting an aircraft, a novel provides sets of problems and gives us practice in navigating in the social domain. Such practice should then–we argue–promote transfer of these skills to the real world. We call this the Social-Improvement Hypothesis. This theory has been explicated in detail elsewhere (Mar & Oatley, in press; cf. Keen 2007; Zunshine 2006). Here we focus on empirical investigations of this idea. Since one of the less understood actors in the social domain is our own self, we also argue that reading fiction, which often focuses on issues of identity, can help self-understanding. Arguably, self-understanding is an important element in changing ourselves. We call this the Self-Improvement Hypothesis: changes in selfhood can occur as a function of reading certain kinds of fiction.
2. Th e social-improvement hypothesis: Does reading fiction improve social skills? George Eliot (1856/1883) proposed that the principal benefit of art is the extension of our sympathies. Among empirical studies related to the idea that literature promotes sympathy is that of van Peer & Maat (1996), who found effects on readers’ sympathies for different characters in a short story as a function of the point of view from which the story was written. Moreover, Hakemulder (2000) has taken up Eliot’s idea of “sympathies,” by proposing that literary fiction is a “moral laboratory.” He searched the psychological literature and found 54 experimental studies that satisfied criteria of reliability and validity, in which fictional narratives promoted moral development, improved empathy, and changed norms, values, and self-concepts. The potential for reading to influence our empathic abilities appears to exist even at a young age. Flerx, Fidler & Rogers (1976) tested five-year-olds who either had fictional stories read to them, watched fiction films that depicted egalitarian sex roles, or watched films with more traditional non-egalitarian sex roles. As compared to those exposed to the more traditional material, children exposed to the egalitarian material showed more egalitarian responses on tests of stereotypes for women’s occupations immediately after the material was presented. A week later, despite some reduction, the effect persisted. These results indicate an improved capacity to empathize with a marginalized group, and we regard this kind of study
Chapter 9. Effects of reading on knowledge, social abilities, and selfhood
as an instance of persuasion by means of narrative (e.g., Green & Brock 2005). In a related experiment using adults (Green 2004), it was shown that individuals who had more experience with the content of a story (i.e., homosexuality, fraternities) reported more engagement and consequently greater shifts in attitude toward those ideas presented in the story. Such studies demonstrate the likely interaction between the creation of empathy for a group through narrative fiction and the capacity for empathy with a group in a narrative fiction based on past experience. The question whether effects of the kinds mentioned above are unique to narrative fiction or whether they also apply to the reading of other types of texts remains. Hakemulder (2000, 2001, see also this volume) provided a possible answer with his experiments using Dutch university students. His hypothesis was that fiction encourages readers to take on the roles of characters in stories, and this makes them more empathetic. Expository non-fiction, of course, lacks such characters. Students were asked to read either a chapter of a novel about the difficult life of an Algerian woman or an essay on the general problem of women’s rights in Algeria. As compared with those who read the essay, those who read the fictional piece said they would be less likely to accept current Algerian norms for relationships between men and women. In another study, Hakemulder found this same decreased tolerance for current norms in students who read the fiction piece under instructions to mentally project themselves into the situation, as compared with those asked to mark the structure of the text with a pencil instead. This follow-up study rules out the possibility that simple text differences are the pivotal variable, and supports the idea that it is our imaginative projection of the self into the described situations that is key. Projecting ourselves into the minds of actual others–inferring their desires, beliefs, and emotions–is known as possessing a theory-of-mind (Astington, Harris & Olson 1988), specifically, the simulation-theory account (for a strong view see Heal, 1998). We explored the idea that this social cognitive process is employed during the comprehension of stories by examining the neuropsychological evidence for this overlap. If the process of story comprehension calls on a process of social cognition then it would be expected that both would draw upon the same areas of the brain. Both the neuroimaging and the neuropsychological (i.e., patient) literatures confirm this. Of the five brain regions consistently associated with narrative processing, four are also part of what is known as the social cognitive network (Mar 2004, cf. Frith & Frith 2003, Saxe & Wexler 2005). Recently, Buckner & Carroll (2007) observed that a network of brain regions appears to be common to a number of different tasks, including theory-of-mind, spatial navigation, autobiographical memory and future planning. While they hypothesized that this
Raymond Mar, Maja Djikic & Keith Oatley
core network was responsible for self-projection, which they believe underlies all these processes, they presented no systematic empirical evidence for the existence of this network. To remedy this situation, Spreng, Mar and Kim (under revision) performed quantitative meta-analyses for these processes (except future planning, for which too few studies exist) and examined how the results of each overlapped. We found ample evidence that a core network contributing to these processes does exist; a number of brain structures were commonly implicated across the different meta-analyses, indicating that these diverse processes share a neural substrate. Moreover, some of these brain regions also overlap with those used for narrative comprehension (Mar 2004). Such findings support the idea that self-projection could explain the link between empathy and the reading of narrative fiction. Whereas the behavioral studies by Hakemulder (2000, 2001) on the role of self-projection have employed the presentation of short texts, our own approach to this question has been to make use of the art to examine how life-time exposure to different genres of text impact empathic abilities. We created a revised version of the art that allowed us to distinguish exposure to narrative fiction from exposure to expository non-fiction. (The small number of items for each genre of fiction unfortunately precludes any analysis based on different types of fiction; this is a question for future research.) Our studies employing this measure, and undergraduate students in Toronto, have indicated that lifetime exposure to fiction does appear related to important social outcomes. In an initial investigation, scores on the art were correlated with performance on two separate social ability tasks (Mar, Oatley, Hirsh, dela Paz & Peterson 2006). Exposure to narrative fiction was positively associated with empathic ability, whereas exposure to expository non-fiction was negatively associated with empathy. Importantly, through the use of partial correlations, we determined that these associations could not be attributed to differences in age, experience with English, and general intelligence. In a follow-up study, we were able to replicate this original finding and also explore possible mediating variables. Using a bootstrapped multiple mediation analysis, we demonstrated that the tendency to imagine oneself as part of a narrative (i.e., self-projection) partially mediated the relation between exposure to narrative fiction and empathic performance, even after considering the role of Openness to Experience, the most relevant Big Five personality trait (Mar, Oatley & Peterson, in preparation). Because the direct effect between fiction and empathy remained statistically significant in our mediation analysis, after taking into account narrative engagement and trait Openness, it is possible that some other factor acting in conjunction with self-projection is also playing a role. We hypothesize that this factor may be practice in understanding social interactions, a skill that could transfer from the reading context to the real social world. In another study, we found that students randomly assigned to read a short story perform better on a subsequent
Chapter 9. Effects of reading on knowledge, social abilities, and selfhood
measure of social reasoning than those assigned to read an essay (Mar 2007). This difference, importantly, does not arise with respect to a non-social measure of analytical reasoning. Taken together, these studies from our own group and others have provided evidence that indicates that the reading of narrative fiction plays a role in developing social expertise. Practice at understanding the fictional social worlds represented by narrative appears to improve our empathic abilities. This research has also illuminated a likely mechanism–projection of the self into the narrative–that is partly responsible for this relation. Important areas of future research include developing a more complete understanding of what this form of self-projection entails, how it is achieved on a neural basis, and also what other variables aside from self-projection can help us understand this relation between fiction and empathy.
3. Th e self-improvement hypothesis: Can reading fiction help change the self? Self-Improvement by reading can be thought of as a branch of bibliotherapy, although with the reading material being literary fiction rather than the usual self-help texts. Narratives are persuasive, and the morals embedded in them are able to change ideas individuals have about the world (Green & Brock 2005). Hakemulder (2000), for example, found that reading a short story about an adulterous love affair, by either Chekhov or Beattie, made men change their attitudes toward adultery in what may be described as a more ethically defensible direction, but only when these stories described a negative outcome for the women involved. Readers, therefore, can adopt the morals implicitly represented in a literary text, and in this way be seen as improving themselves. The process need not be conscious, given that modeling of the ideas presented in narratives (Green & Brock 2005), while requiring an active and imaginative mind, does not require explicit deliberation. Experiences with the morals of stories may not always represent what we would consider self-improvement however, as the possibility exists that readers may choose to model morally murkier aspects of narratives as well. A wealth of literature employs themes of moral ambiguity. These stories are often the most interesting ones, dealing as they do with complex issues that slip the bonds of easy answers. In addition to persuasion, whereby readers report changes in attitudes and beliefs that relate directly to the content of a text, other readers have found that there are consequences of reading that are more dramatic and wide-ranging: changes in their sense of self. Sabine & Sabine (1983) interviewed 1,843 library users as a part of the “Books That Made the Difference” project. They found that their inter viewees considered the books they read to be powerful instigators of self-change.
Raymond Mar, Maja Djikic & Keith Oatley
Ross (1999) found that 60% of readers who read for pleasure (N = 194) found reading to be a personally transforming experience. While lovers of literature may report a profound change in their sense of self as a result of their reading experiences, it is hard not to be skeptical. After all, personality is often defined by its stability, and while it can change across the lifespan (Roberts, Walton & Viechtbauer 2006), this change is likely to be gradual with a diverse number of causes. Perhaps avid readers incorrectly believe their transformation resulted from reading, whereas the true cause lies in other life experiences not tied to experiences with fiction. Perhaps readers inhabit story characters so thoroughly that they think of themselves (incorrectly and temporarily) to be more like these fictional persons. Perhaps their definition of transformation is so broad and vague that it includes any change in opinion or outlook. Perhaps other texts, not only literature, would have as transforming an effect. And even if their self-assessment was correct, who is to say that those literary works that affected them would also affect others? In order to examine the contribution of literary texts to personality change, Djikic, Oatley, Zoeterman, and Peterson (in press) brought 166 undergraduates into a laboratory, and gave them a battery of questionnaires that included a measure of personality traits (the Big Five Inventory; John & Srivastava 1999) and a measure of current emotional state (including ratings of happiness, sadness, boredom, anger, and contentment, among others). Participants were then assigned to one of two conditions. Those in the “Art” condition were given a short story by Chekhov to read, entitled “The Lady with a Toy Dog” (1899). In the “Control” condition participants were given a control text, a rewritten version of the story in a documentary format of a courtroom report of supposed divorced proceedings. A great deal of effort was made to ensure that the story and the transcript were nearly identical save for the form. The control text had all the content of Chekhov’s short story, was exactly the same length, and was of equivalent reading difficulty. Moreover, after reading both texts, participants reported that the court report was just as interesting as the Chekhov story, but not as artistic. After they had read either the Chekhov story or the control text, participants were again given a battery of questionnaires, including the same personality and emotion measures administered initially. A sensitive index of personality change was created such that each post-score was regressed on the pre-score, and the absolute distances were summed to create a composite of personality trait change across all five traits for each individual. The results showed that personality trait change for the participants in the Art condition was significantly greater than the change for the participants in the Control condition. Further analyses revealed that this change in personality was mediated by the emotions that participants experienced while reading.
Chapter 9. Effects of reading on knowledge, social abilities, and selfhood
While it might seem surprising, this study demonstrates that turn-of-thecentury prose by Chekhov can make university undergraduates experience and report themselves as more different than those who read a documentary–style text with the same content, complexity and potential to garner reader interest. It shows that reading literary art can have an effect even on non-avid readers, and that you do not have to be a booklover for reading to transform you. We hypothesize that the effect involves a softening of what are usually the rather rigid boundaries of our self-schemas. By projecting ourselves into fictional stories and the minds of fictional characters, we open ourselves up to greater possibilities for who we may become. It is important for us to stress that participants did not show a collective change in the same direction: not all of them became more extraverted, or open, or conscientious, for example. In other words, they were not persuaded by a moral embedded in a story. Rather, each reader experienced a unique fluctuation in their entire personality profile. Reading Chekhov induced changes in their sense of self–perhaps temporary–such that they experienced themselves not as different in some way prescribed by the story, but as different in a direction toward discovering their own selves. Whether this effect can also be realized with other sorts of fiction has yet to be investigated. Is it possible that, over months and years of reading, we could sum and consolidate such small, and perhaps temporary, changes of the kind we have found here to create movements in the development of selfhood? Our finding with Chekhov’s story prompts us toward believing the claims by avid readers that their favorite literary works have transformed their lives and changed their personalities. We might even start to think of literature in particular, and art in general, as functionally related to human personality development. Might we perhaps take this functionality as a clue to the longevity and persistence of art across millennia of human civilization? 4. Conclusion Although approaching literature by way of empirical study is sometimes seen as reductive, we argue this is not the case. As Willie van Peer has shown in his own career, it is possible to make systematic inquiry into the qualities of literary art and its influence without diminishing the value of fictional literature. Just as an attraction to stories seems to be intrinsically human, so is a curiosity and wonder about the world and the objects in it. Our love of literature and our curiosity about it do not lie in opposition, but are part of the same whole in much the same way our hearts and our minds happily co-exist. But more than that, in our own bodies, our heart could not exist without our mind and vice versa. Although we would not
Raymond Mar, Maja Djikic & Keith Oatley
go so far as to say the same holds true for our love and curiosity about literature, we do feel that the two exist in a mutually beneficial relationship. Our love for literature drives our curiosity, and our curiosity constantly reveals new wonders of literature that serve to magnify our devotion and admiration.
References Astington, J.W., Harris, P.L. & Olson, D.R. (eds). 1988. Developing Theories of Mind. Cambridge: CUP. Buckner, R.L. & Carroll, D.C. 2007. Trends in Cognitive Science 11: 49–57. Djikic, M., Oatley, K., Zoeterman, S., & Oatley, K. In press. On ‘Being Moved’ by art: How reading fiction transforms the self. Creativity Research Journal XX: XX–XX. Echols, L.D., West, R.F., Stanovich, K.E. & Zehr, K.S. 1996. Using children‘s literacy activities to predict growth in verbal cognitive skills: A longitudinal investigation. Journal of Educational Psychology 88: 296–304. Eliot, G. 1883. The natural history of German life: Riehl. In The Works of George Eliot. Standard Edition: Essays.188–236. Edinburgh: Blackwood. (Originally published 1856). Ericsson, K.A. 1990. Theoretical issues in the study of exceptional performance. In Lines of Thinking: Reflections on the Psychology of Rhought, Vol. 2. Skills, Emotion, Creative Processes, Individual Differences and Teaching Thinking, K.J. Gilhooly, M.T.J. Keane, R.H. Logie & G. Erdos (eds), 5–28. Chichester: Wiley. Flerx, V.C., Fidler, D.S. & Rogers, R.W. 1976. Sex role stereotypes: Developmental aspects and early intervention. Child Development 47: 998–1007. Frith, U. & Frith, C.D. 2003. Development and neurophysiology of mentalizing. Philosophical Transactions of the Royal Society of London, Series B 358: 459–473. Green, M.C. 2004. Transportation into narrative worlds: The role of prior knowledge and perceived realism. Discourse Processes 38: 247–266. Green, M.C. & Brock, T.C. 2005. Persuasiveness of narratives. In Persuasion: Psychological Insights and Perspectives, 2nd edn, T.C. Brock & M.C. Green (eds), 117–142. Thousand Oaks CA: Sage. Hakemulder, J.F. 2000. The Moral Laboratory: Experiments Examining the Effects of Reading Literature on Social Perception and Moral Self-concept. Amsterdam: John Benjamins. Hakemulder, J.F. 2001. How to make alle Menschen Brüder: Literature in a multicultural and multiform society. In The Psychology and Sociology of Literature: In Honor of Elrud Ibsch, D. Schram & G. Steen (eds), 225–242. Amsterdam: John Benjamins. Heal, J. 1998. Co-cognition and off-line simulation: Two ways of understanding the simulation approach. Mind and Language 13: 477–498. John, O.P. & Srivastava, S. 1999. The Big Five Trait taxonomy: History, measurement, and theore tical perspectives. In Handbook of Personality: Theory and Research, 2nd edn, L.A. Pervin & O.P. John (eds), 102–138. New York NY: Guilford Press. Keen, S. 2007. Empathy and the Novel. Oxford: OUP. Mar, R.A. 2004. The neuropsychology of narrative: Story comprehension, story production and their interrelation. Neuropsychologia 42: 1414–1434. Mar, R.A. 2007. Simulation-based Theories of Narrative Comprehension: Evidence and implications. PhD dissertation, University of Toronto.
Chapter 9. Effects of reading on knowledge, social abilities, and selfhood
Mar, R.A. & Oatley, K. In press. The function of fiction is the abstraction and simulation of social experience. Perspectives on Psychological Science XX: XX–XX. Mar, R.A., Oatley, K., Hirsh, J., dela Paz, J., & Peterson, J.B. 2006. Bookworms versus nerds: Exposure to fiction versus non-fiction, divergent associations with social ability, and the simulation of fictional social worlds. Journal of Research in Personality 40: 694–712. Mar, R.A., Oatley, K. & Peterson, J.B. In preparation. Exploring the link between reading fiction and empathy: Mechanisms and outcomes. Oatley, K. 1999. Why fiction may be twice as true as fact: Fiction as cognitive and emotional simulation. Review of General Psychology 3: 101–117. Powell, B. 2002. Writing and the Origins of Greek literature. Cambridge: CUP. Roberts, B.W., Walton, K.E. & Viechtbauer, W. 2006. Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin 132: 3–27. Ross, C.S. 1999. Finding without seeking: The information encounter in the context of reading for pleasure. Information Processing and Management 35: 783–799. Ross, P.E. 2006. The expert mind. Scientific American 295(2): 64–71. Sabine, G. & Sabine, P. 1983. Books That Made the Difference. Hamden CN: Library Professional Publications. Saxe, R. & Wexler, A. 2005. Making sense of another mind: The role of the right temporoparietal junction. Neuropsychologia 43: 1391–1399. Siddiqui, S., West, R.F., & Stanovich, K.E. 1998. The influence of print exposure on syllogistic reasoning and knowledge of mental-state verbs. Scientific Studies of Reading 2: 81–96. Spreng, R.N., Mar, R.A. & Kim, A.S. N. Under revision. The common neural basis of autobiographical memory, navigation, theory of mind and the default mode: A quantitative metaanalysis. Journal of Cognitive Neuroscience. Stanovich, K. 1993. Does reading make you smarter? Literacy and the development of verbal intelligence. In Advances in Child Development and Behavior, Vol. 24, H. Reese (ed.), 133–180. New York NY: Academic Press. Stanovich, K.E. & West, R.F. 1989. Exposure to print and orthographic processing. Reading Research Quarterly 24: 402–433. Stanovich, K.E., West, R.F. & Harrison, M.R. 1995. Knowledge growth and maintenance across the life span: The role of print exposure. Developmental Psychology 31: 811–826. Van Peer, W. 1986. Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm. Van Peer, W. 2007. Introduction to foregrounding: A state of the art. Language and Literature 16: 99–104. Van Peer, W. & Maat, H.P. 1996. Perspectivation and sympathy: Effects of narrative point of view. In Empirical Approaches to Literature and Aesthetics. Advances in Discourse Processes, Vol. 52., R.J. Kreuz & M.S. MacNealy (eds), 143–154. Westport CT: Ablex. Wagner, R.K. & Stanovich, K.E. 1996. Expertise in reading. In The Road to Excellence: The Acquisition of Expert Performance in the Arts and Sciences, Sports, and Games, K.A. Ericsson (ed.), 189–225. Hillsdale NJ: Lawrence Erlbaum Associates. West, R.F., Stanovich, K.E. & Mitchell, H.R. 1993. Reading in the real world and its correlates. Reading Research Quarterly 28: 35–50. Zunshine, L. 2006. Why we Read Fiction: Theory of Mind and the Novel. Columbus OH: Ohio State University Press.
chapter 10
Imagining what could happen Effects of taking the role of a character on social cognition Jèmeljan Hakemulder In a series of experiments, Hakemulder (2000) showed that identifying with a story character representing a particular outgroup affects readers’ beliefs about that outgroup. The present contribution describes theories concerning the uses of imagining ourselves in the shoes of story characters. Second, it discusses the research concerning the fundamental processes that may underlie the effects of such imagination. And third, it presents the results of two experiments. The studies of Hakemulder (2000) were replicated, using a different story, examining the effects on not only university students (as in the previous studies) but also on high school students. Again it was shown that role-taking may be responsible for the effects of narratives on social cognition. Keywords: transportation, absorption, narrative effects, social cognition, theory of mind
1. Enter Said and Khaled Seldom do we have a chance to see terrorists in another capacity than in the typical video messages that they leave behind for their family, friends, and enemies – armed to the teeth, agitatedly reciting religious verses or political slogans. In the movie Paradise Now (Abu-Assad 2005), we meet two childhood friends, Said and Khaled, both chosen to be martyrs, and sent into Israel as suicide bombers. In the first twelve minutes, however, unaware of their destiny, we see how they try to make a living in poverty-stricken Palestine, how they spend their free time (shopping, eating, talking politics, making fun, falling in love), and, then, finally, how they are informed that they are to die. As a spectator, you feel you really get to know these two men. Very little information is given about their inner lives, but you do imagine what it would
Jèmeljan Hakemulder
be like to live in the occupied territories, to be ‘honored’ with such a lethal task, and what it would be like to realize that your life is definitely over. It is as if the director invites you to use your imagination in this way. For example, very soon after Said has been told about his duty, you see a strikingly long-lasting (23 seconds), silent medium close-up shot of him. Soon you start forming your own ideas about what this man is thinking, probably using your conception of what you yourself would feel and think. Constructing such a ‘theory of mind’ may enhance a sense of understanding. This is not to propose, by the way, that you will change your moral judgment about terrorism. What I would like to argue here is that narratives like these can affect what viewers consider plausible in a psychological sense. Rather than being aware of the bare fact that some Palestinians do decide to become suicide bombers, spectators now have some idea about how someone comes to the decision to commit such a horrific act of desperation, and how it is to leave your loved ones behind to run around among your enemies with bombs attached to your skin. As a result, the absolutely unimaginable becomes imaginable. Terrorist video messages may generate outright disbelief and stimulate most Westerners to ban such men and women to the domain of all that is inhuman or evil. Paradise Now may elaborate spectators’ conception of what human responses may be like under the circumstances of the Palestinian-Israeli conflict. It seems an easy way to access a stranger’s mind: watching a movie, or reading stories. We do not need to learn foreign languages, get vaccines and visas, make strenuous journeys, nor risk getting ourselves in any life-threatening situations. Narratives allow us to get to know how other people live their lives, all in the comfort of our own homes. Or at least, we think we do. Who knows whether our inferences about Said approach how people in Said’s situation feel and think? Our attributions may be incorrect. The point is that those who talk about the ‘Axis of Evil’ probably have different narratives feeding their ‘theory of mind’ than that of Abu-Assad. The central idea of this contribution is that stories (that is, the plots of novels, movies, soap operas, etc.) are an important source for memory structures or
. When talking about a ‘theory of mind’ I mean the conceptions we form about people’s (and our own) attitudes, feelings, thought, and motives (cf. ‘social cognition’, or ‘attributions’). This should be distinguished from notions of ‘theory of mind’ that refer to an ability to understand other people correctly, as well as definitions that refer to the actual attitudes, feelings etc. that people have at a certain moment. . President George W. Bush’s term, in his State of the Union Address, January 29, 2002, referring to governments that he suspected of supporting terrorism.
Chapter 10. Imagining what could happen
schemata. It is one subgroup of schemata that is particularly relevant here, scripts or scenarios that we use in social cognition (cf. Schank & Abelson 1995). These scenarios may be activated when we try to imagine what it must be like to be in someone else’s shoes. In other words, from all the individual narratives we read, the movies that we see, the stories that we tell each other, we abstract to more general beliefs about what probable sequences of events are. Here we concentrate on readers’ or viewers’ theory of mind, that is, their beliefs about plausible responses to certain situations, about people’s emotions, thoughts, motives; in short, readers’ or viewers’ theory of mind. I will focus on the effects of narratives on the perception of outgroups, that is, social or ethnic groups other than the ones the readers or viewers consider themselves part of. The aim is to explore the practical uses of imagining ourselves in the position of story characters.
2. Imagining scenarios It is a common notion: imagining what it would be like to be in certain unknown situations is a useful tool to acquire knowledge. In several disciplines (e.g., physics, mathematics, law), thought experiments are considered an important vehicle for inquiry, asking hypothetical questions concerning what might happen, or what might have happened in a given situation, and given certain a priori knowledge about the world. Developmental psychologists such as Kohlberg (1969) propose that being able to imagine the consequences of our actions for others is essential for children’s moral development. For this, too, they need to be able to imagine scenarios. The Dutch historian Huizinga (1938) discusses the playful aspect of culture: it is in imagined ‘as if ’ situations, or play, that culture arises (1938: 45). Also outside academia there are numerous situations in which imagining scenarios is assumed to be of practical use. In drama therapy, participants are asked to play out roles in order to relive and subsequently overcome traumatic experiences, or to deal with daily-life problems (cf. Day 2002). In management workshops roleplaying is a tool to practice interactions with colleagues. Furthermore, computer games are used more and more often in education, for example, to explore possible responses to the intricacies of all kinds of real-life situations. In the military, finally, war games are considered a mean to examine the consequences of certain tactics, and to find ways to avoid as many casualties as possible (Ritterfeld & Weber 2006; Lee & Peng 2006). Similarly, speculations about the effects of literature often refer to readers’ involvement in imaginary role-plays (see for an overview Hakemulder 2000). Frank Palmer (1992), for instance, examines what type of knowledge we can
Jèmeljan Hakemulder
expect readers to acquire from reading literature. He argues that it should be defined as “knowledge by acquaintance of what an experience is like,” (218) a type of knowledge that, as Palmer argues, is fundamental to moral reasoning. Richard Rorty (1989) suggests that literature helps us imagine what it must be like to be someone else, a process that may result, he says, in solidarity with those people that the characters represent. “[Solidarity] is created by increasing our sensitivity to the particular details of the pain and humiliation of other unfamiliar sorts of people. Such increased sensitivity makes it more difficult to marginalize people different from ourselves by thinking, ‘they do not feel it as we would.’ ” (xvi) Finally, Willie Van Peer (1996) proposes that literature allows readers to escape the confinement of their own personal opinions, beliefs, views, and habits. Literature instigates them to imagine themselves in a different world. Through concrete portrayals of ‘the other’ the reading experience turns strangers into equals. In its inconspicuous and indirect way, literature feeds its readers’ imagination, which, Van Peer suggests, may have contributed throughout history to major developments (71). There seems to be a rare consensus among literary theorists. I know of no theorist who proposes that reading narrative fiction does not involve an imaginary role-play, and that it does not affects readers’ perception of the world outside the text. But what do we actually know about what happens when we mentally imagine ourselves in the shoes of someone else? For this we briefly turn to what psychologists found out about this.
3. Role-playing One line of research that may be relevant here is that of role-playing studies (e.g., Janis & King 1954). In a number of experiments participants are asked to play an assigned role. For example, they are asked to represent some unpopular position in a debate. As a result, such positions were shown to become more acceptable to the participants. In another example (Janis & Mann 1965), participants are asked to play the role of a patient who is told he or she has lung cancer. Such role-playing proved to be an amazingly effective way to help people stop smoking. The factors that researchers assume are responsible for these effects are the impact that bodily experience has: actually sitting in a chair yourself, with a (actor) physician in front of you at his desk, telling you that you do not have long to live, has a more powerful impact than just witnessing others do the role play. Second, the role-play makes relevant information more prominent in participants’ minds. It may well be that most of us already know that smoking causes lung cancer, but such knowledge is powerfully activated during a role-play. Finally, in terms of selfpersuasion, ideas that we generate ourselves are found to be much more effective
Chapter 10. Imagining what could happen
than information provided by someone else. When participants actively think of arguments in support of some position, they will consider these more valuable than other arguments (Zimbardo & Leippe 1991). How do these role-playing studies relate to the effects of reading? Can we assume that reading narrative texts involves some kind of imaginary role-play with similar results as those found for actual role-play? In a study by Gregory et al. (1982) it was shown that the mere exposure to scenarios is enough to affect beliefs, more in particular, participants’ likelihood estimates for the events described in those scenarios. Both positive and negative scenarios were shown to increase such ratings. Participants were asked to listen to a tape describing a scenario in the second person how someone gets arrested for shoplifting. In another example, participants heard a text about someone winning a prize. In each case participants considered it more likely that these events could happen to them personally than the control group did. There may be some resemblance between role-playing and reading. Readers may be strongly involved in the adventures of story characters. They may identify with them to such a degree that they feel part of the actions and forget the world outside the text. But reading is obviously not quite the same as role-playing. It is stretching things too far to suggest that the results of the role-playing studies are directly relevant to the effects of literature on readers. Also, the texts used in the studies by Gregory et al. (1982) differ in many ways from literary texts. These simple experimenter-generated texts were written to enable control of potentially intervening variables; in a full-blown literary text, however, there are so many factors at play (e.g., multiple characters and story lines) that the effects of reading such narratives are hard to predict. Therefore, it remains unclear whether literary ‘scenarios’ affect readers’ ideas about likely sequences of events. Green & Brock (2000), however, conducted experiments using longer and original texts, showing that readers’ transportation into a narrative instigates a mechanism that results in belief changes consistent with the narrative. Participants with high levels of transportation were more likely to generalize story events to the world outside the text. For example, a narrative describing a murder in a mall increased their estimates of the frequency of murders taking place in the US. Since it may be that readers’ beliefs about murder frequencies caused transportation rather than the other way around, the researchers attempted to manipulate levels of transportation experimentally by instructing some readers to put themselves in the shoes of the story characters, while distracting others from such responses to the narrative. Results of the instruction to enhance transportation were unsuccessful in their studies (experiments 2 and 3). However, in their final experiment, a distraction task did result in a lower level of transportation. Thus the researchers were able to establish a causal effect of transportation on story-consistent beliefs.
Jèmeljan Hakemulder
Hakemulder (2000), too, attempted to establish the effect of role-taking during the reading of a narrative text on likelihood estimates. In one study, participants read either a story about an Algerian woman, or an essay concerning the position of women in fundamentalist Islamic countries. The main character in the story returns to her home country, Algeria, after having lived for several years in France. Soon she discovers why she had left the country in the first place: she is constantly harassed and insulted by boys and men. Since the essay does not offer any direct opportunity to identify with a particular character, it could be argued that its readers are unlikely to be involved in any role-taking. After reading the texts, participants’ perception of the target group, Algerian women, was assessed using a number of items on which they were to indicate the likelihood of certain behavior or attitudes to occur among Algerian women. As expected, it was found that the story was more effective in coloring participants’ imagination than the essay. For instance, compared to the essay group and the control group, story readers considered it less likely that women in Algeria fully accept the relations between men and women. In a second study, the hypothesis that this effect was due to role-taking was further explored by using an instruction variable. Participants were randomly assigned to either a control group or one of two story groups. The same story was used, as in the first experiment. One group was asked to read the story and imagine as much as possible that they were in the position of the story character (cf. Bourg’s 1996 empathy-building instruction). In the second story group, participants were asked to mark the structure of the text using a pen or pencil (the diversion instruction). While role-taking behavior was boosted in the first group, the second group was distracted from such imaginary role-playing. Results showed that both story groups differed significantly in their beliefs about Algerian women from the control group. Moreover, the role-taking instruction affected beliefs significantly more strongly than the diversion instruction. The findings suggest that readers’ imagining themselves in the position of the story character made the scenario more likely to them, ring more true, or seem more plausible to them than to those who read the story with the distracting task. Potentially, these findings could have a huge impact on the role that narratives could play in stimulating a multi-cultural understanding in our societies. Think of the possible uses of immigrant literature, for instance, in education, to help students understand the immigrant experience (cf. Nussbaum 1997). However, the effects that were found may be, of course, particular to the texts that were used. The significant effects suggest that the uses of these texts may lead to similar effects among a random sample of the same population. However, before we can generalize the conclusions to all narratives about outgroups and to all kinds of reader populations, a lot more work needs to be done.
Chapter 10. Imagining what could happen
In the present contribution it will be explored whether the findings could be replicated, using a different story, and representing a different outgroup. Study 1 examines the effects for undergraduate students. To test the possibilities of using literature in secondary education to enhance multi-cultural understanding, Study 2 tests the hypotheses for high school students.
4. Description of two studies In the present experiments the same procedure was used as in Hakemulder (2000). This time, however, a story was selected describing a situation closer to home. Considering the potential uses of narratives in multicultural societies, it is important to test whether the effects found in previous studies also occur when stories describe social groups that are actually part of readers’ environment. In the previous studies it was examined how Dutch readers would be affected by a story about an Algerian woman. It could be argued that the effects might have been due, in part, to the fact that few of the participants ever met an Algerian, and that therefore they were more likely to be influenced. People may be susceptible to information about a country they know very little about. Therefore we could hypothesize that a story concerning a social group familiar to readers is less likely to affect their beliefs. On the other hand, it could also be argued that it is easier for readers to feel transported in a story about a social group they know more about (cf. Green 2004). Considering the hypothesis central to this paper, that being transported in a story is the factor responsible for changes in social cognition, it may be that a story closer to home is more likely to affect outgroup perception than a story about a remote outgroup. With these considerations in mind a story was selected concerning one of the largest ethnic groups in the Netherlands, the Turks. The book the text was selected from is Mukhtar’s Turkish Gold (2007), which relates the experience of a Turkish girl, Aselya, who falls in love with a Sudanese boy. The section taken from the novel (pages 136–147) describes how her father reacts when he finds out. He is infuriated because his daughter did not inform him about the affair. In addition, he already had another husband for her in mind, back in Turkey. He beats her and locks her up in her room. Family members threaten to kill the lover. The section ends with Aselya’s escape. As in one of the previous studies (Hakemulder 2000), an essay concerning a similar theme was selected (Van der Zee’s (2006) Honor violence in the Netherlands, pages 91–99). Again, the idea was to test how the effects of a narrative would relate to materials that do not allow readers to identify with a character, or to learn ‘through acquaintance what an experience is like.’
Jèmeljan Hakemulder
4.1 Study 1 In the first study, 166 Dutch students (125 female, 38 male, three cases missing) enrolled in a course called ‘Introduction to literary studies’ were asked to participate in an experiment. The mean group age was 20.6 (SD = 2.5). Participants were told that the study concerned “Memory and reading style.” They would be asked to “read the text in a certain way to test whether this would affect their recall.” Participants were randomly assigned to one of four groups: a control group that read an unrelated story; a group that read the essay by Van der Zee; or one of two story groups, with one receiving a role-taking instruction, and the other a diversion instruction. The role-taking instruction was the same as used in Hakemulder (2000). It read “Try to put yourself as much as possible in the position of the main story character. Try to imagine while reading how it would be for you to be in her position. Consider what she sees, what she smells, and what she feels. Try to imagine that all that happens to her, happens to you yourself.” The diversion instruction was meant to distract participants from identifying with the character: they were to mark the structure of the text with a pen or pencil. After reading the text, all participants were given the same task. They were asked to imagine what it would be like to be a Turkish girl growing up in the Netherlands. To assess how the treatments affected their theory of mind, the following items were used: “Would you want to choose your own husband?”, “Would you always obey your parents?”, “Would you consider it normal that your parents would punish you physically?”, “Would you accept the way women are treated?”, “Would you resist unequal rights for men and women?”. These items were to measure story consistent beliefs. The girl in the story does want to choose her own husband; she does not obey her parents; she does not accept her beating as normal; she resists how she is treated by escaping. All questions were answered using a six-point scale, ranging from “absolutely unlikely” to “very plausible”. Six weeks after the experiment all participants were fully debriefed about the purpose and the outcome of the study. 4.2 Results The story group with the role-taking and the group that read the essay show effects in opposite directions (Table 1). Readers who were asked to take the role of the Turkish girl were less likely to accept physical punishments, while essay readers were more likely to. Interestingly, comparing the two story groups we can see that role-taking is a necessary condition for the effects of the narrative. Likelihood estimates of the group that was told to pay attention to the story structure did not differ significantly from those of the control group. The character, Aselya, clearly does not want her father to beat her; she flees with her lover, resisting further punishment. This scenario became more likely to those readers who were asked
Chapter 10. Imagining what could happen
Table 1. Undergraduate students Belief Would want to choose own husband Would always obey parents Normal to be punished physically Accept how women are treated Would resist unequal rights
Control M
SD
Role-taking M
SD
Diversion M
SD
Essay M
SD
4.92 .93 5.11* 1.11 4.80 + .87 3.95* + 1.74 3.75 1.32 3.70 1.48 3.71 1.29 4.10 1.41 2.83 + o 1.30 2.03* + 1.18 2.82x 1.42 4.10*ox 1.15 3.21 1.25 2.86 1.46 2.94 1.25 3.89 1.15 4.12x 1.03 4.31* 1.24 4.15+ 1.18 2.95* +x 1.27
Note: *, x and + indicate couples of mean groups scores that differ significantly, (p < .05). Numbers in bold indicate significant differences with the control group scores.
to identify with Aselya. The essay readers, on the other hand, read about the place of girls in Turkish families and as a result seem to have assumed it the practice of physical punishment to be normal. Undergraduate students are the most thoroughly examined group in psychology. However, it is not the most obvious group to focus on in the present enterprise. When we want to argue that literary education may have an important function in a multicultural society, we need to test the hypotheses for younger age groups. Therefore, an attempt was made to replicate the findings of Study 1 in a high school setting. 4.3 Study 2 In the second experiment, 99 students of Higher General Secondary Education participated. The mean age was some four years younger than that of the participants of Study 1 (M = 16.7, SD = .74). Roughly the same procedures were used. However, a less difficult control text was used, and some minor adjustments were made in the wording of the questionnaire to match the vocabulary of the younger participants. 4.4 Results The results of the second study are summarized in Table 2. Contrary to expectations, the effects on participants’ likelihood estimates do not correspond with
. Higher General Secondary Education, or Hoger Algemeen Voorgezet Onderwijs (havo), prepares students between 12 and 17 for higher vocational training. Thanks to Liesbeth Belloni of Broklede College in Breukelen for her cooperation.
Jèmeljan Hakemulder
Table 2. High school students Belief Would want to choose own husband Would always obey parents Normal to be punished physically Accept how women are treated Would resist unequal rights
Control M
SD
Role-taking M
SD
Diversion M
SD
Essay M
SD
5.65* + .49 3.45 1.32
5.17* .92 3.63 1.47
5.50 1.10 3.73 1.72
4.71+ 1.87 3.27 1.67
2.20 1.44
2.08 1.38
2.00 1.23
2.95 1.73
2.26 1.24 4.50 + 1.15
2.42 1.25 3.79 1.38
2.18 1.26 3.95 1.29
2.95 1.40 3.50 + 1.79
Note: *, + and x indicate couples of mean groups scores that differ significantly, (p < .05). Numbers in bold indicate significant differences with the control group scores.
the story scenario. Both the role-taking group and the essay group considered it significantly less likely that a Turkish girl would want to choose her own husband than the control group. It seems that these role-taking readers drew a different conclusion than the students in the previous study did: since the girl is punished so severely, it might have seemed less likely to them that a girl would take such risks. The essay group behaved as expected. After reading about the customs among Turkish parents to marry off their daughters, the essay readers considered it less likely than the control group that these girls want to choose their own husband. It may be that the older age group of Study 1, most of them probably living on their own, used different personal experiences than the younger group still living with their parents. It is important to note that role-taking caused a different effect here than expected. Nevertheless, it seems, again, that it was a necessary condition for the narrative to affect readers’ likelihood estimates. Where we find a significant difference between the role-taking group and the control group, none were found for the story group with the diversion instruction.
5. Discussion and conclusion The present attempts to replicate the findings of Hakemulder (2000) suggest that taking the role of a character may be responsible for the effects of narratives on outgroup perception. It seems that stories feed readers’ theory of mind, even if the represented group is familiar to them: we registered effects on the perception of Algerian women, as well as Turkish girls growing up in the Netherlands. However,
Chapter 10. Imagining what could happen
different populations exposed to the same story were found to focus on different scenarios they see represented in the narrative. High school students, the participants of Study 2, drew different conclusions from the story than the undergraduate students of Study 1. The results show that determining what “story consistent” beliefs are is not unproblematic. Although the two studies show differences in the items on which effects were established (compare Table 1 and 2), and the effects in Study 2 were the opposite of the expected direction, there is one constant in the results: when we compare the two story groups in both studies, we see that the role-taking instruction consistently leads to effects on social cognition, while the diversion instruction does not. Obviously we cannot claim that role-taking is the sole way to affect social cognition; in the two experiments presented here, contrary to the earlier findings, we did find significant effects for the essay. Is it high time to start using narratives to create multicultural understanding, tolerance, and solidarity? Of course, the results of these few studies do not warrant such bold conclusions. First of all, as explained before, the direction of the effects is unpredictable. Using full-blown stories has the advantage of external validity. But the present studies also illustrate the downside: the loss of control of known and unknown potential variables can lead to unforeseen consequences. The fact that the outcome of these or similar experiments is still uncertain is a major obstacle for large-scale implementation of narrative-based curricula. Second, we do not know what the long-term effects of imaginary role-taking may be, if any. It could be argued that the results only show that reading stories primes certain schemata that are subsequently used in social cognition tasks. On the other hand, it may be that repeated exposure to one and the same type of plot does have long-term effects. Cultivation research (cf. Gerbner & Gross 1976) shows a correlation between exposure to crime stories and beliefs about the omnipresence of crime in society. These effects were also established experimentally (Bryant et al. 1981), showing that a heavy diet of action and adventure causes higher perceived likelihood of victimization. Similarly, the more of the same type of stories we read about outgroups, the more we may be affected by them, also in the longer run. So, after reading several stories about Turkish girls who resist their parents’ judgments, readers might be more likely to change their ideas about Turkish girls. A third problem that arises is the question whether role-taking is a common aspect of the reading experience. In the experiments one particular aspect of reading was boosted, using a role-taking instruction; since few books and movies are accompanied with such instructions, it seems doubtful whether the findings can be generalized to reading situations in society at large. However, it does seem that readers often feel part of the fictional world, as if they are taking the position of one
Jèmeljan Hakemulder
of the characters in the story. van der Bolt & Tellegen (1993), for instance, showed that high levels of absorption while reading is typical of adult readers rather than of children, as some have suggested. Finally, we should consider the question of whether we are dealing with typically literary effects. ‘Literariness’ is rarely defined in terms of emersion into fictional worlds, or identification with story characters. It seems that these responses are considered more characteristic of readers of popular fiction. However, several theorists have argued that narrative comprehension in general requires readers to formulate some theory of mind (cf. Palmer 2004; Zunshine 2006). Miall & Kuiken (1995) using their Literary Response Questionnaire show significant correlations between absorption and ‘Empathy’, ‘Insight’, and ‘Imagery vividness’, while a negative correlation was found between absorption and ‘Rejection of literary values.’ This suggests that in the experience of the readers, being transported into a narrative does go very well together with the appreciation of literature. There are still a great number of issues that need to be resolved. In addition to the problems mentioned above concerning the generalizability of the findings, future research may try to pinpoint the role of fictionality in the effects of narratives. It may be that these effects do not depend on whether they are fictional or not (cf. Green & Brock 2004). Narratives seem to have the power to make events seem more likely to happen. Even when we are aware that the events are very unlikely to occur in our own lives, at an involuntary level we sometimes feel they may very well happen. Cantor (2004) found that many of her respondents experience such effects as a result of frightening movies they saw years ago. A typical example is that of the spectators of Jaws: they know that in some waters the presence of a Great White is very unlikely. But still the thought that it is there, waiting to grab their legs, is irresistible to them. Research in this field is inconclusive about the role of fictionality. It seems likely, however, that in many cases it does not hinder the effects. The question remains, when is it effective, and when not. The research on the persuasiveness of narratives in general is similarly inconclusive. One meta-analysis shows that narratives are stronger than facts (Reinard 1988), another shows the opposite, favoring facts (Allen & Preiss 1997). It is important to find out more about the role of narratives in, for instance, judgments of social problems (cf. Strange & Leung 1999), especially with some movie productions more rapidly tapping into current affairs than ever before.
. See A.O. Scott. Seeking the truth of war, with ambigious results. International Herald Tribune, October 30, 2007.
Chapter 10. Imagining what could happen
One way to further explore the persuasiveness of narratives and the role of fictionality is to look at other factors that may explain why some fictional narratives are, and others are not effective in changing real-world beliefs. It may be argued that Jaws-like scenarios are, in a way, much more self-relevant, that is, ‘informative’ about life-threatening situations. Therefore such stories may have a longer lasting and stronger impact than stories about ethnic groups. However, it may be that scenario’s about outgroups are self-relevant, admittedly in a less direct way. They might help us understand and predict the behavior of people in our environment, thus satisfying a fundamental need. Future research should look into issues of relevance: is self-relevance essential for the effects of fictional narratives? And what factors determine the level of self-relevance? It seems likely that perceived parallels between readers’ own life and that of the story characters may play a role. Also, stories that touch upon readers’ personal and most central concerns may also be more effective than those that deal with relatively more peripheral issues (Tan 1996). One of the contributions of Willie Van Peer to the empirical study of literature has been a general shift in researchers’ attention, from a psycholinguistic interest in the uses of world knowledge in text comprehension, to a more socially relevant focus on the effects of narratives. As a result, we now understand a little better how the world is reorganized in terms of the work, and the work in term of the world (cf. Goodman 1976: 241). Future developments may encompass the exploration of the specific ways in which literary reading affects role-taking. In such studies one should distinguish three interrelated conceptions: first, as was examined in the present experiments, the influence of narratives on social cognition. Second, reading literature may train role-taking abilities. This notion should be set apart from a third idea, namely that reading narratives enhances empathy. Goldstein and Winner (under review) found that playing a role may indeed enhance theory of mind abilities, but it does not necessarily lead to higher levels of empathy. Comprehension of literary narratives may pose challenges to readers’ theory of mind that exceed by far those of other genres. Because of this aspect of literary texts, it may be that the second notion, the effects of reading on theory of mind abilities, deserves some more attention. Maybe the merging interest of neuropsychology for literature (see Mar 2007, and elsewhere in this volume) may in the future uncover what the exact contribution of literary reading is to such abilities. The relevance of such research may be clear, considering the impact of these abilities on the quality of communication. There are many more questions to be solved. However, if stories do indeed affect our understanding of others, or if readers do become more willing to try to understand people from other cultures, it is certainly worth the trouble to focus our research on these problems. A 13th century poem by Jalal ad-Din Rumi
Jèmeljan Hakemulder
narrates how four travelers get into a fight because they can’t agree on how to spend the one coin they have together. The Persian decides he wants to buy engur; the Arab wants inab, the Turk insists on üzüm, and the Greek demands stafil. Unable and unwilling to communicate, none of them gets what they want. If reading literature stimulates us to reflect on what others think and feel, it may help us, in the end, to understand what they want. And maybe we discover we all want the same – just as the four travelers, who all intended to spend their last coin the same way: on grapes.
References Abu-Assad, H. (director). 2005. Paradise Now. The Netherlands. Allen, M. & Preiss, R.W. 1997. Comparing the persuasiveness of narrative and statistical evidence using meta-analysis. Communication Research Reports, 14(2): 125–131. Bourg, T. 1996. The role of emotion, empathy, and text structure in children’s and adults’ narrative text comprehension. In Empirical Approaches to Literature and Aesthetics, R. Kreutz & M.S. MacNealy (eds), 241–260. Norwood NJ: Ablex. Bryant, J. Carveth, R. & Brown, D. 1981. Television viewing and anxiety: An experimental examination. Journal of Communication 31: 106–119. Cantor, J. 2004. “I’ll never have a clown in my house.” – Why movie horror lives on. Poetics Today 25(2): 283–304. Day, L. 2002. Putting yourself in other people’s shoes. The use of Forum theatre to explore refugee and homeless issues in schools. Journal of Moral Education 31(1): 21–34. Gerbner, G. & Gross, L. 1976. Living with television. The violence profile. Journal of Communication 26: 173–199. Goodman, N. 1976. Languages of Art. Indianapolis IN: Hackett. Goldstein, T. & Winner, E. (under review). Actors as mind-reading experts: Acting training predicts advanced levels of theory of mind but not empathy. Green, M.C. 2004. Transportation into narrative worlds: The role of prior knowledge and perceived realism. Discourse Processes 38(2): 247–266. Green, M. & Brock, T. 2000. The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology 79(5): 701–721. Gregory, W.L., Cialdini, R.B. & Carpenter, K.M. 1982. Self-relevant scenarios as mediators of likelihood estimates and compliance: Does imagining make it so? Journal of Personality and Social Psychology 43: 89–99. Hakemulder, J. 2000. The Moral Laboratory. Experiments Examining the Effects of Reading Literature on Social Perception and Moral Self-concept. Amsterdam: John Benjamins. Huizinga, J. 1938. Homo Ludens. Proeve eener bepaling van het spel-element der cultuur. (Homo Ludens. A Study of the Play Element in Culture). Boston MA: Beacon Press. Janis, I.L. & King, B.T. 1954. The influence of role-playing on opinion change. Journal of Abnormal and Social Psychology 49: 211–218. Janis, I.L. & Mann, K. 1965. Effectiveness of emotional role-playing in modifying smoking habits and attitudes. Journal of Experimental Psychology in Personality 1: 84–90.
Chapter 10. Imagining what could happen
Lee, K.M. & Peng, W. 2006. What do we know about social and psychological effects of computer games? A comprehensive review of the current literature. In Playing Video Games. Motives, Responses and Consequences, P. Vorderer & J. Bryant (eds), 327–345. Mahwah NJ: Lawrence Erlbaum. Mar, R.A. 2007, November. Theory-of-mind and self-projection during narrative comprehension: Neuropsychological evidence. Paper presented at the Theory of Mind and Literature Conference. Purdue University. West Lafayette IN, USA. Miall, D. & Kuiken, D. 1995. Aspects of literary response: A new questionnaire. Research in the Teaching of English 29(1): 37–58. Mukhtar, S. 2007. Turks Goud. (Turkish Gold). Amsterdam: Van Gennep. Nussbaum, M. 1997. Cultivating Humanity. A Classical Defense of Reform in Liberal Education. Cambridge MA: Harvard University Press. Kohlberg, L. 1969. Stage and sequence. The cognitive-developmental approach to socialization. In Handbook of Socialization Theory and Research, D.A. Goslin (ed.), 347–380. New York NY: Rand McNally. Palmer, A. 2004. Fictional Minds. Lincoln, NB: University of Nebraska Press. Palmer, F. 1992. Literature and Moral Understanding. A Philosophical Essay on Ethics, Aesthetics, Education, and Culture. Oxford: Clarendon Press. Reinhard, J. 1988. Empirical study of the persuasive effects of evidence. The status after fifty years of research. Human Communication Research 15: 3–59. Ritterfeld, U. & Weber, R. 2006. Video games for entertainment and education. In (eds), 399–413. Playing Video Games. Motives, Responses and Consequences, P. Vorderer & J. Bryant (eds). Mahwah NJ: Lawrence Erlbaum. Rorty, R. 1989. Contingency, Irony, and Solidarity. Cambridge: CUP. Schank, R.C. & Abelson, R.P. 1995. Knowledge and memory: The real story. In Knowledge and Memory: The Real Story [Advances in Social Cognitions VIII], R. Wyer (ed.), 1–85. Hillsdale NJ: Lawrence Erlbaum. Strange, J. & Leung, C. 1999. How anecdotal accounts in news and in fiction can influence judgments of a social problem’s urgency, causes, and cures. Personality and Social Psychology Bulletin 25: 436–449. Tan, E.S. 1996. Emotion and the Structure of Narrative Film. Film as an Emotion Machine. Mahwah NJ: Lawrence Erlbaum. Van der Bolt, L. & Tellegen, S. 1993. Involvement while reading: An empirical exploration. Imagination, Cognition, and Personality 12: 273–286. Van der Zee, R. 2006. Eerwraak in Nederland. (Honor Violence in the Netherlands). Amsterdam: Houtekiet. Van Peer, W. 1996. Literatuur als list. (Literature as a cunning trick). In Een Beeld van Belezenheid. Over Culturele Geletterdheid. (An Image of Being Well-Read. On Culture and Literary Literacy), R. Soetaert & L. Top. (eds), 75–90. Den Haag: sdu. Zimbardo, P. & Leippe, M. 1991. The Psychology of Attitude Change and Social Influence. New York NY: McGraw-Hill. Zunshine, L. 2006. Why We Read Fiction: Theory of Mind and the Novel. Columbus OH: Ohio State University Press.
part iii
Computers and the Humanities
Computers and the Humanities Introduction The prophecy that computing will transform the nature of literary studies is certainly one that we have heard before, but the widespread use of powerful personal computers in the last few years and the increasing role played by the internet, now makes such a forecast seem to carry more weight (Miall 1995: 199).
This is how David Miall described the influence of computers in the field of literature in 1995. By now, 13 years later, it is possible to assess their importance. They have not replaced human analysts, as some feared in the beginning. On the contrary, they have been an invaluable aid in research: computers have proved to be much more reliable than humans when it comes to counting, sorting and identifying words in large (collections of) texts. In addition, they perform these activities with increasing speed. Depending on the research, it may be now inconceivable to use paper slips and do all the work by hand. Even simple tasks as counting the occurrences of ‘the’ in this short introduction would take some time and would probably yield different results if two people were asked to carry it out. Computers have become part of everyday life, including academic research in the field of the Humanities. Their impact can be reflected, for instance, on the growing number of international associations such as the Association for Computers and the Humanities, the Association for Literary and Linguistic Computing, and journals such as Computers and the Humanities, Literary and Linguistic Computing and Language Learning & Technology. The articles in this section show in practical ways how computers and some of their tools may be used to investigate texts, mostly literary. All of them follow the definition of empirical studies forwarded in the introduction to this volume. The real evidence concerns the text itself, which may be easily investigated under a completely different light with the help of computer tools. It is now possible to focus on a large number of linguistic features, which was highly unlikely in the pre-computational era. Given the objectivity of such studies, they may be replicated. Therefore, the literary investigations reported here favor objective analyses over subjective judgments. In other words, instead of writing on what they believe literary texts . As a matter of fact, there are 87 instances of ‘the’ in this introduction.
Directions in Empirical Literary Studies
to be, the authors of the six chapters in this section resort to concrete evidence of language use. This section starts with computers per se and ends in the Humanities. This means the initial focus lies on the tools that may enable researchers to probe texts in novel ways and produce quite unexpected findings. As readers proceed with the reading, they will notice a shift towards studies that focus more on the language of literary discourse. The opening chapter by Art Graesser and Brent Morgan centers on a text analytical computer tool named Coh-Metrix. In order to show the usefulness of the program, he uses as material a selection of Willie van Peer’s articles published in the 1980’s, 1990’s and 2000’s. The result of such timely idea for this Festschrift is an overview of the changes van Peer’s style has undergone over a period of 20 years. In “Computationally discriminating literary from non-literary texts”, Max Louwerse, Nick Benesh and Bin Zhang also examine common language use, but in their paper there is already a shift towards literary language. The authors report on five studies aimed at characterizing the difference between literature and nonliterature by means of language features. In a provocative way, they end the paper by proposing that literary discourse may be characterized by a specific two-word sequence, which differentiates it from other discursive types. Michael Kimmel’s chapter “Metaphors and software-assisted cognitive stylistics” differs from the previous chapters in that it highlights the use of a computer program for qualitative analyses of literary texts. The author draws on concepts from Cognitive Linguistics to analyze metaphors in one literary work, namely, Henry James’s Turn of the Screw. As the chapter develops, Kimmel also looks into Herman Melville’s Billy Budd and Ian McEwan’s Cement Garden for the sake of comparison. From a general perspective, the chapter covers some important methodological issues regarding computer-assisted metaphor research. The fourth text is more literature-oriented. David L. Hoover analyzes poems by Thomas Stearns Eliot, Robert Lee Frost, Edwin Arlington Robinson, Carl August Sandburg and Wallace Stevens with the help of Iota and Zeta, two measures of textual differences. The procedure adopted in the chapter brings to the forefront the words which characterize the stylistic production of such well-known American poets. Colin Martindale’s chapter entitled “The laws governing the history of poetry” also works with poems, but the author concentrates on the production of British writers. The text provides readers with a comprehensive discussion on novelty in art, especially literature. In a historical approach ranging from 1290 to 1949, the author probes the production of 170 poets to test the approach described in the text.
Computers and the Humanities
In the chapter that closes this section, Bill Louw discusses the concepts of ‘digital collocation’ and ‘semantic prosody’, advocating a change in stylistic methodology – one that favors objectivity and observable data. Additionally, the author proposes the development of a corpus-attested glossary of literary terms in order to systematize the knowledge that has been produced so far. At the end of the text, Louw provides readers with a glimpse of such a glossary by offering sample definitions for ‘insincerity’, ‘irony’, ‘metaphor’, ‘semantic prosody’ and ‘worlds’. The six chapters grouped here show different ways in which computers may be used in the Humanities, which reminds us of Galileo in the sense that: The telescope was invented in 1608 and was initially thought useful in war. Galileo obtained one, improved it a little, and used it to challenge existing ideas about the Solar System. Although a magnificent new technology in itself, the telescope was hardly a scientific tool until Galileo used it to create new knowledge (De Smedt 2002: 99).
It is exactly this idea of generating new thoughts that characterizes the studies in this section. The six of them offer a fresh approach to literature by analyzing empirical data. Computer tools are used here to produce new knowledge in this field, thus enabling new vistas of the familiar. Vander Viana Section Editor
References De Smedt, K. 2002. Some reflections on studies in humanities computing. Literary and Linguistic Computing 17(1): 89–101. Miall, D.S. 1995. Representing and interpreting literature by computer. Yearbook of English Studies 25: 199–212, http://www.ualberta.ca/~dmiall/complit.htm (accessed December 19, 2007).
chapter 11
An automated text analysis Willie Van Peer’s academic contributions Arthur C. Graesser & Brent Morgan During the last 15 years there have been radical advances in computational linguistics that analyze language at the levels of words, sentences, paragraphs, rhetorical structure, world knowledge, pragmatics, and context. Among these advances is Coh-Metrix, a computer tool developed at the University of Memphis. Coh-Metrix (http://cohmetrix.memphis.edu) is a facility on the web that automatically analyzes texts on numerous dimensions of words, syntax, referential cohesion, coherence of the mental models, and genre in addition to standard readability measures. This chapter analyzes some texts of Willie van Peer’s academic contributions with Coh-Metrix, following his appeal early in his career that computers might play an important role in exploring the characteristics of literary and non-literary texts. We tracked characteristics of his writing over two decades to explore how his writing style has changed. Keywords: cohesion, computational linguistics, corpus analysis, syntax, mental models. Coh-Metrix
1. Introduction The authors of this chapter and Willie van Peer have shared similar visions on how to investigate the process of comprehending literature, art, and media. It is important to take an interdisciplinary perspective that integrates (a) a systematic analysis of the aesthetic object, (b) an empirical analysis of the psychological processes of comprehension, and (c) a theoretical grounding in science, philosophy, and other scholarly enterprises. This is the vision that inspired the creation of igel (Internationale Gesellschaft für Empirische Literaturwissenschaft, The Society for the Empirical Study of Literature and Media). The first author of this chapter served as the third president of igel and Willie van Peer, the tenth (see mention in the introduction to this volume).
Arthur C. Graesser & Brent Morgan
The destinies of Art Graesser and Willie van Peer crossed at another important level when two of Willie van Peer’s students (Rolf Zwaan & Max Louwerse) were postdoctoral fellows at the University of Memphis. Rolf Zwaan went on to build a stellar career investigating mental models (Zwaan, Magliano & Graesser 1995; Zwaan & Radvansky 1998) and subsequently a framework of embodied cognition that integrates language, perception, and action (Zwaan & Taylor 2006). Max Louwerse made substantial research contributions on the psychological processing of connectives (Louwerse 2001), corpus-based analyses of discourse markers (Louwerse & Mitchell 2003), features of language that predict genre (Louwerse, McCarthy, McNamara & Graesser 2004), thematics (Louwerse & van Peer 2002), and multichannel communication (Louwerse, Graesser, Lu & Mitchell 2005). Max Louwerse is the eleventh president of igel, following Willie van Peer in the position. There is yet another vision that is shared by Willie van Peer and the authors of this chapter. Automated computer analyzes of texts can be a powerful tool for researchers who pursue empirical analyses of literary and non-literary text comprehension. As van Peer expressed nearly 25 years ago, “Computer-assisted study of literature has by and large concentrated mainly on linguistic characteristics of the text. In general, this approach has led to fruitful insights into matters such as the stylistic traits typical of a particular genre, author, work, period, or tradition” (van Peer 1983). Indeed, there has been a dramatic increase in computer analyses of large text corpora during the last 15 years. This can partly be explained by revolutionary advances in computational linguistics (Jurafsky & Martin 2000), lexicons (Fellbaum 1998; Pennebaker & Francis 1999), discourse processes (Pickering & Garrod 2004; Graesser, Gernsbacher & Goldman 2003), the representation of world knowledge (Lenat 1995; Landauer, McNamara, Dennis & Kintsch 2007), and corpus analyses (Biber, Conrad & Reppen 1998). Millions of words in documents can be quickly accessed and analyzed on thousands of measures in a short amount of time. This vision of van Peer has been directly realized in many of the research projects in the Institute for Intelligent Systems (iis) at the University of Memphis. Researchers in the IIS have developed several computer systems that analyze the English language, such as Coh-Metrix (http://cohmetrix.memphis.edu, Graesser, McNamara, Louwerse & Cai 2004), quaid (Graesser, Cai, Louwerse & Daniel 2006), AutoTutor (Graesser et al., 2004), and iSTART (McNamara, Boonthum, Levinstein & Millis 2007). The Coh-Metrix system is particularly relevant to this chapter because it covers the full spectrum of language, discourse, and world knowledge. In addition to handling over 200 categories and properties of words, there are automated analyses of syntax, referential cohesion, coherence of mental models, and discourse structure. The values on the Coh-Metrix measures can be
Chapter 11. An automated text analysis: Willie Van Peer’s academic contributions
used to investigate the cohesion of the explicit text and the coherence of the mental representation. Our definition of cohesion consists of linguistic characteristics of the explicit text that play some role in connecting ideas in the text. Coherence includes aspects of cohesion that are likely to contribute to the coherence of mental representations. In order to illustrate the value of Coh-Metrix, we analyzed a small corpus of scholarly texts written by Willie van Peer. We analyzed excerpts from two of his articles published in the mid-1980’s (van Peer 1983, 1986), four from the mid1990’s (van Peer 1995, 1996, 1997; van Peer & van der Knaap, 1995), and two from the early 2000’s (van Peer 2001, 2002). These scholarly journal articles or book chapters reported theoretical and empirical articles on literature, language, stylistics, and discourse. This selection of articles concentrated on the most outstanding scholarly outlets at these three points in time. We randomly selected a paragraph from the beginning, middle, and end of each article. The mean number of words sampled from each article was 783 words, ranging from 609 to 1297 words. This is obviously a small corpus, but the modest goal of this chapter is to illustrate the value of Coh-Metrix rather than to perform a rigorous scientific text analysis of Willie van Peer’s writing. How might we expect the scholarly language of van Peer to change over a 20 year time span? Many different trends would be quite plausible, but a few are articulated below. (a) Complexity hypothesis. Language and discourse might become more complex as expertise grows. If so, the content words would become longer (more jargon and rare words), the syntax would be more complex structurally, and the analytical composition of the ideas would have more extensive embedding, recursion, negations, and hypothetical or conditional expressions. Complexity of knowledge is generally expected to increase as a function of growth of expertise (Ericsson, Charness, Feltovich & Hoffman 2006). On the other hand, perhaps an accomplished expert writer is more sensitive to the reader and acquires the skill of communicating in plain language (Kellogg 1994). Many of us have identified the passionate inexperienced writer who tries to impress the reader with erudite jargon, convoluted syntax, and conceptually dense arguments that obfuscate more than illuminate. (b) Oral-literate continuum hypothesis. The language prepared for print is expected to be more compact, literate, and structurally dense than the language of the oral tradition (Biber 1988; Chafe 1982; Tannen 1982). Oral language has shorter words, more pronouns, simpler syntax, and content that is more distributed over sentences (e.g., typically only one new idea per clause or sentence). The language of print is more complex because it is planned ahead of time, and most sentences have multiple new ideas embedded in its noun-phrases and support
Arthur C. Graesser & Brent Morgan
clauses. Perhaps there is a shift from an oral to a literate style as expertise grows. On the other hand, accomplished experts may know how to better communicate with the readers; if so, perhaps the style shifts from the literature to oral end of the continuum because oral language styles are normally easier to comprehend. (c) Cohesion hypothesis. Expertise may promote the composition of texts with more cohesion and coherence. Alternatively, the opposite might be expected because experts might have blind spots on what readers know (i.e., the pathology of expertise) and incorrectly assume a high common ground (shared knowledge) between writer and reader. Cohesion and coherence markers are not needed as much when there is higher domain knowledge and common ground (Clark 1996; McNamara & Kintsch 1996). These possibilities would be reflected in measures of referential cohesion and the coherence of the situation model. It is possible to entertain many other hypotheses, but these three will be considered in the present chapter.
2. Coh-Metrix There are over 60 measures of language and discourse in the Coh-Metrix version (v. 2.0) that is available to the public on a website. Coh-Metrix was designed to move beyond standard readability formulas, such as Flesch-Kincaid Grade Level (Klare, 1974–1975), which rely exclusively on word length and sentence length. For example, the formula below shows the scoring of Flesch-Kincaid Reading Ease. Words refers to mean number of words per sentence and syllables refers to mean number of syllables per word. Flesch Reading Ease = 206.835 – 1.015 x Words – 84.6 x Syllables Coh-Metrix provides deeper measures of language and discourse that are no doubt important to comprehension. Some refer to characteristics of individual words, as has been achieved in many other computer facilities, such as WordNet (Fellbaum 1998) and Linguistic Inquiry Word Count (Pennebaker & Francis 1999). However, the majority of the Coh-Metrix measures are more processingintensive algorithms that analyze syntax, referential cohesion, semantic cohesion, text structure, and the coherence of the referential mental models of what the text is about. There are over 700 measures in the internal Coh-Metrix system, whereas approximately 60 of these are available in the public. 2.1 Words Coh-Metrix measures words on a large number of characteristics that can be viewed in the help facilities on the website. There are measures of word frequency in the English language, which is based on the celex lexicon (Baayen, Piepenbrock &
Chapter 11. An automated text analysis: Willie Van Peer’s academic contributions
van Rijn 1993) and other similar lexicons. Coh-Metrix identifies different classes of content words (e.g., nouns, main verbs, adjectives) and function words (e.g., prepositions, articles), based on standard part-of-speech categories that are accepted in the computational linguistic community. The incidence of each word class (i.e., relative frequency) is computed as the number of occurrences per 1000 words. An incidence score is necessary for comparing texts of different sizes. Several word measures are directly relevant to cohesion, coherence, and comprehension difficulty. In particular, there are word classes that have the special function of connecting clauses and other constituents in the text (Halliday & Hasan 1976; Louwerse 2001; Sanders & Noordman 2000). The categories of connectives in Coh-Metrix include additive (also, in addition), temporal (then, after, during), causal (because, so), and logical operators (therefore, if, and, not, or). The word indices include negations (not, n’t) that span different levels of constituent structure and various conditional expressions (if, given). Negations, conditional expressions, and negative connectives are predicted to be affiliated with complex conceptualizations and rhetorical structures, such as counterfactuals, hypothetical worlds, multiple perspectives, qualifications, hedges, and argumentation. A higher incidence of these words should therefore predict text complexity. 2.2 Syntax Coh-Metrix analyzes sentence syntax with the assistance of a syntactic parser developed by Charniak (2000). The parser assigns part-of-speech categories to words and syntactic tree structures to sentences. Coh-Metrix has several indices of syntactic complexity, but only three of them are reported in this article. The mean number of modifiers per noun-phrase is a measure of the complexity of referencing expressions. For example, the seductively beautiful actress is a complex noun-phrase with 3 modifiers of the head noun actress. The number of words before the main verb of the main clause is a measure of syntactic complexity because it places a burden on the working memory of the comprehender (Graesser, Cai, Louwerse & Daniel 2006). For example, in the sentence As the painter observed the sunset over the Mississippi River, the tugboat slowly moved upstream, the main verb of the main clause is moved and there are 13 words that precede it. Syntactic similarity of sentences is a measure of the extent to which pairs of sentences in a text (i.e., adjacent sentences or all possible sentences in a paragraph) have similar syntactic tree structures. A higher score means that the writer uses similar syntactic constructions from sentence to sentence and lower syntactic diversity. 2.3 Referential and semantic cohesion Referential cohesion occurs when a noun, pronoun, or noun-phrase refers to another constituent in the text. For example, in the sentence As the woman approached
Arthur C. Graesser & Brent Morgan
the child, she grabbed him immediately, the word she refers to the word woman and him refers to the child. The resolution of pronoun referents is quite challenging to accurately accomplish by computers (Jurafsky & Martin 2000; Lappin & Leass 1994) and sometimes difficult for humans. A text with a high incidence of pronouns may suffer from problems of cohesion and coherence. It is easier to establish referential connections when the words have root morphemes (lemmas) that overlap. One form of co-reference that has been extensively studied is argument overlap (Kintsch & van Dijk 1978). This occurs when a noun, pronoun, or noun-phrase in one sentence is a co-referent of a noun, pronoun, or noun-phrase in another sentence. The word “argument” is used in a special sense in this context, namely, it is a noun phrase that plays a thematic role in the sentence (agent, object, recipient, time, place) and is contrasted with predicates (i.e., main verbs, adjectives) in propositional representations. The argument overlap measure of Coh-Metrix currently considers exact matches of arguments between two sentences. The value of this measure varies from 0 to 1 and is the proportion of adjacent sentence pairs that share a common argument in the form of an exact match. Another form of co-reference is stem overlap, where a noun in one sentence has a similar lemma to a content word in another sentence. For example, consider the two sentences: The college students danced in the nightclub. The dances became more erotic as the night went on. Danced and dances have common stems, so there is stem overlap even though one is a main verb and the other a noun. In addition to referential cohesion measures, Coh-Metrix can measure the extent to which the content of sentences or paragraphs is similar semantically or conceptually. Latent Semantic Analysis (lsa) is one important method of computing similarity because it considers implicit knowledge. lsa is a statistical technique for representing world knowledge, based on a large corpus of texts with approximately 10 million words. The central intuition is that the meaning of a word is captured by the company of other words that surround it in naturalistic documents. Two words have similarity in meaning to the extent that they share similar surrounding words (i.e., encyclopedia type knowledge). lsa uses a statistical technique called singular value decomposition to condense a very large corpus of texts to 100–500 statistical dimensions (Landauer, McNamara, Dennis & Kintsch 2007). The conceptual similarity between any two text excerpts (e.g., word, clause, sentence, text) is computed as the geometric cosine between the values and weighted dimensions of the two text excerpts. The value of the cosine varies from 0 to 1. lsa-based coherence was measured in two ways from the standpoint of the present study: (1) lsa similarity between adjacent sentences and (2) lsa similarity between all possible pairs of sentences in a paragraph. Lexical diversity is a simple measure of computing the redundancy of the content in a text. The lexical diversity measure in Coh-Metrix we will use in this
Chapter 11. An automated text analysis: Willie Van Peer’s academic contributions
chapter is the type-token ratio score for content words. This is the number of unique words in a text (i.e., types) divided by the overall number of words (i.e., tokens) in the text. A low value means there is a large amount of redundancy in the content words of a text, whereas a high value means that each content word is used once or very few times. Redundancy is inversely related to the type-token ratio. 2.4 Mental model dimensions Many aspects of a text can contribute to the mental model (sometimes called the situation model), which is the referential content or microworld of what a text is about (Graesser, Singer & Trabasso 1994; Kintsch 1998). Text comprehension researchers have investigated at least five dimensions of the situational model (Zwaan, Magliano & Graesser 1995; Zwaan & Radvansky 1998): causation, intentionality, time, space, and protagonists. A break in coherence occurs when there is a discontinuity on one or more of these mental model dimensions. Whenever there are such discontinuities in conceptual content, it is important to have connectives, transitional phrases, adverbs, or other signaling devices which convey to readers that there is a discontinuity. We refer to these different forms of signaling as particles. Coherence is facilitated by particles that clarify and stitch together the actions, goals, events, and states conveyed in the text. Coh-Metrix 2.0 analyzes the mental model dimension on causation, intentionality, space, and time, but not protagonists. There are many measures of the cohesion or coherence (in this context the two terms can be used interchangeably) of the mental model, but the present study concentrated on the following measures: the incidence of all connectives, the incidence of causal connectives, causal coherence, and temporal coherence. We have already addressed connectives and causal connectives, which reflect increases in cohesion and coherence. For causal cohesion, Coh-Metrix computes the ratio of cohesion particles to relevant referential content (i.e., main verbs that signal state changes, events, actions, and processes, as opposed to states). The ratio metric is essentially a conditionalized incidence of cohesion particles: given the occurrence of relevant content (such as clauses with events or actions, but not states), what is the density of particles that stitch together the clauses? For example, the actions and events that occur in stories, scripts, and common procedures are very relevant content whereas the cohesion particles would include diagnostic connectives (because, so), adverbs (consequently), and prepositions (to, by). The referential content for causation information includes various classes of events that are identified by change-of-state verbs and other relevant classes of verbs in WordNet (Fellbaum 1998). In the case of temporal cohesion, Coh-Metrix computes the uniformity of the sequence of main verbs with respect to tense and aspect. Cohesion increases as a function of the sequential
Arthur C. Graesser & Brent Morgan
uniformity of verbs with respect to tense and aspect. Texts that hop around with respect to tense and aspect would have lower temporal cohesion, as in the case of flashbacks and flash forwards. 2.5 Text structure and genre Two other measures are worthy of mention but will not be operationally defined in this chapter. First, topic sentencehood is the proportion of sentences that are computed as topic sentences in a paragraph, based on various algorithms in computational linguistics. Information texts, which normally occur in print, would be expected to have higher topic sentencehood scores than narrative discourse in the oral tradition. Second, genre purity is the extent to which a text fits a particular specific genre, such as narrative, science, versus social studies; a low score would be assigned to a text that is an amalgamation of multiple genres.
3. The scholarly texts of Willie van Peer Now that the Coh-Metrix system has been described, we are ready to turn to an analysis of the small corpus of scholarly writings of Willie van Peer. Does his writing shift to greater complexity, literateness, and cohesion over the 20 year span? Or does it become more simple, oral, and lower in cohesion? Table 1 presents the mean scores for the various Coh-Metrix measures as a function of the years of his writings (1980’s, 1990’s, versus 2000’s). The measures are segregated into different clusters, such as simple measures of texts, word-level measures, syntax, referential and semantic cohesion/coherence, mental model coherence, and discourse structure. The final column specifies whether each measure increased over the years or decreased. The direction was defined when comparing the magnitudes of the means during the 1980’s versus the later years (i.e., 1980 < 1990 and 1980 < 2000). These decision rules yielded a consistent pattern for all measures except for the first one, namely average words per sentence. That is, the mean in 1980 was greater than 1990, but lower than 2000; the mean sentence length was 25.2 in the 1980’s and 25.5 when averaging the 1990’s and 2000’s, which is quantitatively similar and not different statistically. Therefore, Willie van Peer has not seemed to lengthen or shorten his sentences over the years. It is important to note, however, that we did not perform inferential statistics to compare these means because, as noted earlier, this is really an illustration of the use of CohMetrix rather than a thorough analysis of Willie van Peer’s writing. We will leave it to others to conduct an exhaustive analysis of Willie van Peer’s texts.
Chapter 11. An automated text analysis: Willie Van Peer’s academic contributions
Table 1. Means of Coh-Metrix measures of Willie van Peer’s writing
1980’s
1990’s
2000’s
Trend
Average words per sentence
25.2
23.9
27.0
Average syllables per word Flesch-Kincaid Reading Ease
1.72 35.5
1.67 41.3
1.64 40.8
Decrease Increase
2.13
2.22
2.27
Increase
SIMPLE MEASURES OF TEXTS
WORD LEVEL Logarithm of frequency of content words Incidence of pronouns
Pronoun per noun-phrase ratio
12.2 .12
29.5 .17
38.2 .18
Increase Increase
Incidence of all connectives
69.6
79.0
76.2
Increase
Incidence of all logical operators
25.6
46.3
38.1
Increase
3.9
14.0
11.0
Increase
(and + if + or + conditional + negation) Incidence of negations SYNTAX Average number of modifiers per noun phrase
Words before main verb of main clause in sentences Sentence syntax similarity
1.09
.94
.87
Decrease
6.9 .08
6.3 .07
5.5 .06
Decrease Decrease
.91
.84
.90
Decrease
.94
.87
.90
Decrease
REFERENTIAL, SEMANTIC, AND CONCEPTUAL COHESION AND COHERENCE Argument overlap of adjacent Sentences Stem overlap of adjacent sentences
lsa cosine of adjacent sentence to sentence
lsa cosine of all sentence pairs in paragraph Type-token ratio of all content words
.29
.21
.27
Decrease
.27 45.6
.20 43.9
.26 42.5
Decrease Decrease
19.0
12.3
11.6
Decrease
MENTAL MODEL COHERENCE Incidence of causal connectives
Causal cohesion: Causal particles divided by causal verbs Temporal cohesion: Tense and aspect repetition scores DISCOURSE STRUCTURE Topic sentencehood
2.5 .61
1.5 .56
.7 .50
Decrease Decrease
.25
.21
.20
Decrease
The results revealed that the trends in van Peer’s writings very much depended on the level of text analysis. Consider first the word level. The Flesch-Kincaid reading ease index has increased over the years, so he has apparently attempted to make it easier for his readers to understand his writing. The increase in reading ease can be attributed to having shorter words (i.e., fewer syllables per word) rather than
Arthur C. Graesser & Brent Morgan
shorter sentence length because the latter stayed constant, as already mentioned. His writing also has had a higher incidence of pronouns and a higher proportion of noun-phrases that are filled with pronouns. Pronouns are, of course, diagnostic of the oral language rather than the literate end of the oral-literate continuum. The story of sentence syntax follows a similar trend of greater simplicity over the years. He has been considerate on the reader by having fewer words in nounphrases and fewer words before the main verb of the main clause in the sentences. This syntactic profile eases the load on the reader’s working memory, so once again Willie van Peer appears to be sympathetic to designing the language for the reader. At this point, we might conclude that van Peer has evolved into a champion of clear writing for the common person, as opposed to a stilted academic that obfuscates the message and befuddles the reader. However, the landscape of trends shifts as we move to deeper levels of text processing, beyond the word and the syntax of individual sentences. We find that his language became more analytically dense with respect to the negation words and the full suite of Boolean logical operators (e.g., and, or, if, not). Thus, it would appear his razor-sharp logical mind intensified in his writings over the years. From the standpoint of writing clarity, one penalty of academic expertise is that the reader must suffer with dense logical arguments. Another consequence of the growth of Willie van Peer’s expertise is that cohesion and coherence ended up decreasing over the years. According to the measures in the lower half of Table 1, there was a decrease in the measures of (1) argument overlap, stem overlap, and lsa scores of adjacent sentences, (2) lsa scores of all possible sentence pairs within a paragraph, (3) incidence of causal connectives, (4) causal and temporal cohesion, and (5) topic sentencehood. The greater use of pronouns is another indicator that cohesion tended to decrease because pronouns sometimes pose difficulties when their referents cannot be resolved. The similarity of syntax across sentences also tended to decrease. These decreases cannot be readily attributed to the content being more difficult or covering disparate ideas because the type-token ratio of content words tended to decrease (greater redundancy) rather than increase. The trend also cannot be attributed to his using fewer connectives because the incidence of connectives increased rather than decreased. In summary, these data suggest that discourse cohesion and coherence drifted lower over the years for Willie van Peer. What might we conclude when we revisit the complexity, oral-literate continuum, and coherence hypotheses? There is a mixed picture from the standpoint of complexity. His words and syntax became simpler, whereas his logical arguments became more complex. The pictures are a bit clearer for the other hypotheses. His texts shifted from the literate to the oral end of the continuum over the years, perhaps in an effort to make his work easier to understand and more engaging.
Chapter 11. An automated text analysis: Willie Van Peer’s academic contributions
However, there was decrease in discourse cohesion and coherence. A decrease in discourse coherence/cohesion presents greater comprehension difficulties for the low-knowledge reader, but improves comprehension for the medium to highknowledge reader (McNamara & Kintsch 1996) by virtue of inviting inferences, deeper processing, and engagement. From this perspective, Willie van Peer has come to tune his writing to the typical readers of his work.
4. Closing comments This chapter has reported some changes in Willie van Peer’s writing over twenty years, but we close by addressing his presentations and writing compared with his peers. Van Peer’s peers at igel would probably agree with us in making the following three claims. First, his discourse is indeed extremely easy to follow and as engaging as a minstrel storyteller. This is in marked contrast to the ubiquitous counterparts who lapse into dense, abstract, turgid, ineffectual prose. Second, van Peer has always been attuned to the controversial issues of literary comprehension from the standpoint of theory and empirical studies: he has never sidestepped an angry controversy or tiptoed evasively into a safe corner. And third, his razorsharp analytical mind has always shined in a good nasty argument that fans the flames. Willie van Peer reminds us all that a good analytical mind is not always incompatible with the most extreme emotional temperament.
Acknowledgements The research on Coh-Metrix was supported by the Institute of Education Sciences (R3056020018–02). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of ies.
References Baayen, R.H., Piepenbrock, R. & van Rijn, H. (eds). 1993. The CELEX Lexical Database (CD-ROM). University of Pennsylvania, Philadelphia PA: Linguistic Data Consortium. Biber, D. 1988. Variations Across Speech and Writing. Cambridge: CUP. Biber, D., Conrad, S. & Reppen, R. 1998. Corpus Linguistics: Investigating Language Structure and Use. Cambridge: CUP. Chafe, W.L. 1982. Integration and involvement in speaking, writing, and oral literature. In Spoken and Written Language, D. Tannen (ed.), 35–54. Norwood NJ: Ablex.
Arthur C. Graesser & Brent Morgan Charniak, E. 2000. A maximum-entropy-inspired parser. In Proceedings of the First Conference on North American Chapter of the Association for Computational Linguistics, 132–139. San Francisco CA: Morgan Kaufmann. Clark, H.H. 1996. Using Language. Cambridge: CUP. Ericsson, K.A., Charness, N., Feltovich, P.J. & Hoffman, R.R. (eds). 2006. The Cambridge Handbook of Expertise and Expert Performance. Cambridge: CUP. Fellbaum, C. (ed.). 1998. WordNet: An Electronic Lexical Database. Cambridge MA: The MIT Press. Graesser, A.C., Cai, Z., Louwerse, M. & Daniel, F. 2006. Question understanding aid (QUAID): A web facility that helps survey methodologists improve the comprehensibility of questions. Public Opinion Quarterly 70: 3–22. Graesser, A.C., Gernsbacher, M.A. & Goldman, S. (eds). 2003. Handbook of Discourse Processes. Mahwah NJ: Lawrence Erlbaum. Graesser, A.C., Lu, S., Jackson, G.T., Mitchell, H., Ventura, M., Olney, A. & Louwerse, M.M. 2004. AutoTutor: A tutor with dialogue in natural language. Behavioral Research Methods, Instruments, and Computers 36: 180–193. Graesser, A.C., McNamara, D.S., Louwerse, M.M. & Cai, Z. 2004. Coh-Metrix: Analysis of text on cohesion and language. Behavior Research Methods, Instruments, and Computers 36: 193–202. Graesser, A.C., Singer, M. & Trabasso, T. 1994. Constructing inferences during narrative text comprehension. Psychological Review 101: 371–395. Halliday, M.A.K. & Hasan, R. 1976. Cohesion in English. London: Longman. Jurafsky, D. & Martin, J.H. 2000. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition. Upper Saddle River NJ: Prentice-Hall. Kellogg, R.T. 1994. The Psychology of Writing. Oxford: OUP. Kintsch, W. 1998. Comprehension: A Paradigm for Cognition. Cambridge: CUP. Kintsch, W. & van Dijk, T.A. 1978. Toward a model of text comprehension and production. Psychological Review 85: 363–394. Klare, G.R. 1974–1975. Assessing readability. Reading Research Quarterly 10: 62–102. Landauer, T., McNamara, D., Dennis, S. & Kintsch, W. (eds). 2007. Handbook of Latent Semantic Analysis. Mahwah NJ: Lawrence Erlbaum. Lappin, S. & Leass, H.J. 1994. An algorithm for pronominal coreference resolution. Computational Linguistics 20: 535–561. Lenat, D.B. 1995. CYC: A large-scale investment in knowledge infrastructure. Communications of the ACM 38: 33–38. Louwerse, M.M., Graesser, A.C., Lu, S. & Mitchell, H.H. 2005. Social cues in animated conversational agents. Applied Cognitive Psychology 19: 693–704. Louwerse, M.M. 2001. An analytic and cognitive parameterization of coherence relations. Cognitive Linguistics 12: 291–315. Louwerse, M.M., McCarthy, P.M., McNamara, D.S. & Graesser, A.C. 2004. Variation in language and cohesion across written and spoken registers. In Proceedings of the 26th Annual Meeting of the Cognitive Science Society, K. Forbus, D. Gentner & T. Regier (eds), 843–848. Mahwah NJ: Lawrence Erlbaum. Louwerse, M.M. & Mitchell, H.H. 2003. Toward a taxonomy of a set of discourse markers in dialog: A theoretical and computational linguistic account. Discourse Processes 35: 199–239.
Chapter 11. An automated text analysis: Willie Van Peer’s academic contributions
Louwerse, M.M. & van Peer, W. (eds). 2002. Thematics: Interdisciplinary Studies. Amsterdam: John Benjamins. McNamara, D.S., Boonthum, C., Levinstein, I. & Millis, K. 2007. Evaluating self-explanations in iSTART: Comparing word-based and LSA-based algorithms. In Handbook of Latent Semantic Analysis, T. Landauer, D.S. McNamara, S. Dennis & W. Kintsch (eds), 227–242. Mahwah NJ: Lawrence Erlbaum. McNamara, D.S. & Kintsch, W. 1996. Learning from text: Effects of prior knowledge and text coherence. Discourse Processes 22: 247–287. Pennebaker, J.W. & Francis, M.E. 1999. Linguistic Inquiry and Word Count (LIWC). Mahwah NJ: Lawrence Erlbaum. Pickering, M.J. & Garrod, S. 2004. Toward a mechanistic psychology of dialogue. Brain and Behavioral Sciences 27: 169–190. Sanders, T.J.M. & Noordman, L.G.M. 2000. The role of coherence relations and their linguistic markers in text processing. Discourse Processes 29: 37–60. Tannen, D. 1982. The oral/literate continuum in discourse. In Spoken and written language, D. Tannen (ed.), 1–16. Norwood NJ: Ablex. Van Peer, W. 1983. Poetic style and reader response: An exercise in empirical semics. Journal of Literary Semantics 12: 3–18. Van Peer, W. 1986. Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm. Van Peer, W. 1995. Literature, imagination, and human rights. Philosophy and Literature 19: 276–291. Van Peer, W. 1996. Canon formation: Ideology or aesthetic quality? The British Journal of Aesthetics 36: 97–108. Van Peer, W. 1997. ‘High’ / ‘Low’ cultural products and their social functions. Empirical Studies of the Arts 15: 29–39. Van Peer, W. 2001. Justice in perspective. In New Perspectives on Narrative Perspective, W. van Peer & S. Chatman (eds), 325–338. Albany NY: Suny Press. Van Peer, W. 2002. Where do literary themes come from? In Thematics: Interdisciplinary Studies, M. Louwerse & W. van Peer (eds), 253–263. Amsterdam: John Benjamins. Van Peer, W. & van der Knaap, E. 1995. (In)compatible interpretations? Contesting readings of ‘The Turn of the Screw’. Comparative Literature 4: 692–710. Zwaan, R.A., Magliano, J.P. & Graesser, A.C. 1995. Dimensions of situation-model construction in narrative comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition 21: 386–397. Zwaan, R.A. & Radvansky, G.A. 1998. Situation models in language comprehension and memory. Psychological Bulletin 123: 162–185. Zwaan, R.A. & Taylor, L.J. 2006. Seeing, acting, understanding: Motorresonance in language comprehension. Journal of Experimental Psychology: General 135: 1–11.
chapter 12
Computationally discriminating literary from non-literary texts Max Louwerse, Nick Benesh & Bin Zhang Three computational linguistic methods are presented to discriminate literary from non-literary texts. In the first study, a hierarchical clustering technique of results obtained from Latent Semantic Analysis showed a clustering of literary versus non-literary texts. The second study used the frequencies of shared bigrams across the text, resulting in a 100% correct classification of literary versus non-literary texts. The third study used unigrams yielding a 94% correct classification into literary versus non-literary texts. The final two studies using a larger sample of texts showed that the high classification performance cannot be attributed to specific texts. These findings provide evidence that distinguishing literature from non-literature can be done with high accuracy and with relatively simple computational linguistic techniques. Keywords: computational linguistics, stylistics, genre, bigram analysis, latent semantic analysis, classification techniques
1. Introduction It does not take much effort for those who visited him to recall his office on Muntstraat 4 in Utrecht, on the second floor immediately left from the squeaky stairs. Coffee in a plastic cup from the machine at the end of the corridor; inside his office two desks covered with piles of papers, the ones that had not quite made it into the binders that filled the shelves on the wall. Radiators turned high, the door slightly open as a silent invitation to researchers in the field of empirical studies of literature. At the time, the first author was a student in Literary Studies and research assistant of Willie van Peer. Literary studies students study literature. If somebody asks what constitutes as literature one ought to reply with terms like “aesthetic”, “deconstruction”, “fictionality”, “defamiliarization”, “syuzhet”, “foregrounding” and “literariness” (Jefferson & Robey 1986). The first author recited these same terms when his research assistants, the
Max Louwerse, Nick Benesh & Bin Zhang
co-authors of this paper, asked that question. The problem with this answer is that it may be possible to identify prototypical examples of literary language from literature (Fabb 2002; Fowler 1996), but whether literary texts overall are linguistically different from non-literary texts is a question that has not been satisfactorily answered. In fact, very little is known about the language that distinguishes literature from non-literature and whether computational linguistic techniques can offer answers (see Louw 1993; Sinclair 2004; Stubbs 2005). Perhaps the answers these computational linguistic techniques provide are “utterly naïve” (van Peer 1989: 302) because the fundamental literary ingredients are reduced to lower levels of linguistic organization. That is, “no level of (mathematical) sophistication is able to overcome the problem that the processes of meaning constitution have been eliminated before the analysis is undertaken” (van Peer 1989: 302). Indeed, the text and its meaning should be considered (Sinclair 2004) and one should refrain from taking a text to pieces (Sinclair 1966). The aim of the current chapter was to do exactly that: use computational linguistic techniques that take literary and non-literary texts to pieces to provide, what van Peer (1989: 302) called, “utterly naïve” answers. More specifically, this paper asks the question of whether distinctions can be made between literary from non-literary texts by using three computational linguistic methods: 1) a higherorder co-occurrence analysis of the words in all texts; 2) a computation across all texts by considering the frequencies of bigrams, or word pairs, 3) an account per text of the frequencies of specific words or types of words. The findings presented in these five studies provide evidence that distinguishing literature from non-literature can be done with high accuracy and with relatively simple computational linguistic techniques.
2. Study 1 One way to investigate the semantic content of texts is by looking at the semantic neighbors of words. Sentences like “The researcher worked on foregrounding in literary texts”, “The researcher worked on stylistics in literary texts”, and “The researcher worked on the empirical studies of literary texts” suggest that “foregrounding”, “stylistics” and “empirical studies” have semantic content in common. Sentences or paragraphs however often do not have the same semantic context, resulting in a sparsity problem. Higher-order relationships between words can then be the solution, by considering the neighbors of the neighbors of the neighbors etc. of words. Latent Semantic Analysis (lsa) is a statistical technique that estimates the semantic relations between words, sentences, paragraphs or texts by
Chapter 12. Computationally discriminating literary from non-literary texts
considering these higher-order co-occurrences of words. Meaning is captured by mapping initially meaningless words into a continuous high-dimensional semantic space, which more or less simulates cognition (Landauer 2002). More specifically, a first-order process associates stimuli (words) and the contexts they occur in (documents). Stimuli are paired based on their contiguity or co-occurrence. These local associations are next transformed by means of Singular Value Decomposition into a small number of dimensions (typically 300) yielding more unified knowledge representations by removing noise. Because it considers associations between concepts, represented by words, lsa could be seen as a theory of knowledge representation, induction and language acquisition (Landauer & Dumais 1997; Landauer, McNamara, Dennis & Kintsch 2007; for a comprehensive introduction to lsa, see Kintsch 2002; see also Louwerse 2004, Louwerse & van Peer 2006; in press, for a demonstration of how lsa can be extremely helpful in the analysis of literary texts). 2.1 Materials In this first study, we compared the semantic content in literary and non-literary texts using lsa in order to determine whether semantic content allows for clustering of texts in at least these two groups. The materials used were nine literary and seven non-literary texts. Texts were considered literary if they appeared in Zane’s book (2007) Top Ten List, whereby he polled 125 British and American top authors asking them to pick the top 10 books of literature of all time. Nine of the top ten books were made electronically available; one could not (Marcel Proust’s In Search of Lost Time). Non-literary texts were chosen from a set of corpora that was used in previous studies (Crossley & Louwerse 2007). An overview of these 16 texts is presented in Table 1. Note that numbers of texts in the two conditions slightly differed, but that analyses in this study and further studies are not sensitive to unequal cell sizes. The nine literary texts differed in size (e.g., War and Peace and The Great Gatsby), genre (e.g., Letters of Anton Chekhov, Hamlet, Anna Karenina), original language versus English translations (e.g., Madame Bovary, Middlemarch) and year of publication (e.g., Hamlet, Lolita). The seven non-literary texts differ in formality (e.g., Santa Barbara Corpus, Trains), in written versus spoken register (e.g., Wall Street Journal, Edinburgh MapTask Corpus) and modality (e.g., Switchboard, Santa Barbara Corpus, New York Times). The Edinburgh MapTask Corpus (Human Communication Research Centre, 1993) is a task-based corpus that is the linguistic product of a cooperative task involving two participants. Instruction givers have a marked route on
Max Louwerse, Nick Benesh & Bin Zhang
Table 1. Texts and corpora used in Studies 1–3 Author
Title
Year
Word count
Genre
L. Tolstoy Anna Karenina W. Shakespeare Hamlet A. Chekhov Letters of Anton Chekhov V. Nabokov Lolita G. Flaubert Madame Bovary G. Eliot Middlemarch M. Twain Adventures of Huckleberry Finn F. Fitzgerald The Great Gatsby L. Tolstoy War and Peace
1877 1601 1888 1955 1856 1871 1884
358,689 32,756 118,639 116,752 117,554 326,025 116,547
Literature Literature Literature Literature Literature Literature Literature
1925 1869
50,110 572,535
Literature Literature
Memphis MapTask 2008 343,050 Edinburgh MapTask 1991 2,731,809 New York Times 1996 29,560,931 Santa Barbara 2000 464,932 Switchboard 1997 3,702,166 TRAINS 1995 84,149 Wall Street Journal 1996 17,956,625
Dialogue (task-based) Dialogue (task-based) Newspaper Dialogue (informal) Dialogue (telephone) Dialogue (text-to-text) Newspaper
their map and give directions to the instruction followers who have no route. The maps are not identical, which elicits unscripted problem solving dialog. The Memphis Multimodal MapTask Corpus (Memphis MapTask Corpus; Louwerse, Bard, Steedman & Graesser 2004) is similar to the Edinburgh Corpus except that different maps were chosen and the setup focused on multimodal communication. The corpus was selected here as a comparison to the Edinburgh MapTask Corpus, with a subset of only 32 (out of the 256 dialogues) being used. The trains Corpus (Allen & Heeman 1995) is based on the routing and scheduling of freight trains. The corpus shares with MapTask its basis as a task-based corpus, but it is more temporal and directional in nature than the spatial MapTask Corpus. Two nontask-based dialogue corpora are the Switchboard Corpus (Godfrey & Holliman 1993) and the Santa Barbara Corpus (Du Bois, Chafe, Meyer & Thompson 2000). The Switchboard Corpus is a collection of 2,400 two-sided random topic telephone conversations taken from 543 speakers from all areas of the United States. The Santa Barbara Corpus is a collection of natural speech recordings taken from people across the United States. Finally, the New York Times and Wall Street Journal corpora consist of all articles from those newspapers from 1996 (Graff 1996). The texts included
Chapter 12. Computationally discriminating literary from non-literary texts
in this study are extremely diverse. Because of the diversity of the literary as well as the non-literary texts an accurate classification of these texts into literary and nonliterary on the basis of semantic content is far from obvious. 2.2 Results For lsa analyses, a ‘knowledge base’ is needed in the form of an lsa space. The semantic relation between two or more words, sentences, paragraphs, or – in the current case – texts can be computed using this knowledge base. The general Touchstone Applied Science Associates (tasa) Corpus is commonly used to create such a space and is only used for that purpose in this study. The tasa Corpus consists of approximately 10 million words of unmarked high school level English texts on Language Arts, Health, Home Economics, Industrial Arts, Science, Social Studies, and Business. This corpus is divided into 37,600 documents, averaging 166 words per document, and is considered one of the benchmark corpora in computational linguistics, because it approximates the language familiarity of a college level student (Landauer & Dumais 1997). The matrix of 16x16 cosine values representing the semantic similarities between the texts was submitted to a hierarchical clustering analysis, using a between-groups linkage clustering method and a squared Euclidean distance measure. The dendogram resulting from this analysis is presented in Figure 1. Figure 1 shows that by taking the semantic content of the texts, lsa is able to discriminate between two groups of texts, literary and non-literary. Within the non-literary group, the task-oriented dialogues cluster together (Edinburgh MapTask Corpus and Memphis MapTask Corpus, as well as Trains), followed by a grouping of the other dialogues. Next, the two newspaper corpora clustered together. The selected dialogues of the two Maptask corpora considerably differed in size (the selected Edinburgh Corpus being almost eight times larger). The fact Anna Karenina Madame Bovary War and Peace Middlemarch Hamlet Letters Chekhov Lolita Adv. of Huckleberry Finn The Great Gatsby New York Times Wall Street J. Santa Barbara Switchboard Memphis MapTask Edinburgh MapTask TRAINS
Figure 1. Hierarchical clustering of lsa cosine values between all 16 texts.
Max Louwerse, Nick Benesh & Bin Zhang
that the two clustered closely together therefore suggests that lsa is not sensitive to corpus size. Zane’s (2007) top 3 literary works (Anna Karenina, Madame Bovary, War and Peace) clustered together, with Middlemarch being added to this cluster. Another cluster that emerges is that of two novels that can be considered “the canon of American self-images and values” (Rowe 1988: 13) (Adventures of Huckleberry Finn, The Great Gatsby). Those texts with a different style, because of genre (Letters of Anton Chekhov, Hamlet) or the narrating style (Lolita; see Green 1987) also formed a cluster. The main result of this analysis, however, is that an analysis based on semantic relations across all lexical items of the texts allows for a categorization of literary and non-literary texts. Identical results as those in the hierarchical clustering method were obtained using an alscal Multidimensional Scaling (mds) representation algorithm, with the matrix of lsa cosine values being transformed into a matrix of Squared Euclidean distances. mds does not result in a hierarchical clustering, but the stimulus coordinates, followed an identical pattern as the hierarchical clustering algorithm. The advantage of these stimulus coordinates is that they can later be used for correlational analyses. The mds fitting of the data was satisfactory (Kruskal’s stress 1 = .25; R2 = .84) with a one-dimensional scaling. The advantage of an lsa analysis is that all semantic content is taken into account when clustering the text. The drawback is that because a higher-order co-occurrence technique is used, it is difficult to determine what linguistic information is accountable for these clusters. As discussed in Crossley & Louwerse (2007) and Louwerse, Lewis and Wu (in press), a bigram analysis can account for this problem. The same 16 texts were classified using a bigram analysis in Study 2.
3. Study 2 Bigrams are combinations of two words occurring in a corpus. A text consisting of a sentence “John loves Mary and Mary John” consists of the bigrams “John loves”, “loves Mary”, “Mary and”, “and Mary”, “Mary John”. One could take the frequency of these bigrams and compare them with the frequencies of bigrams shared in other texts. If the bigram “and Mary” is high in Text 1 and 2, and significantly lower in Text 3 and 4, two groups of texts can be identified. Bigrams (or rather n-grams) have been used in a number of language models such as determining the probability of a sequence of words, speech recognition models, spelling correction, machine translation systems, and optical character recognizers (Jurafsky & Martin 2000; Manning & Schutze 1999). For instance, Crossley & Louwerse (2007) used bigrams for register classification of spoken and
Chapter 12. Computationally discriminating literary from non-literary texts
written corpora, and Louwerse, Lewis & Wu (in press) used bigrams to categorize the genres in Shakespeare’s plays. Study 2 tested whether bigrams also allow for a classification of texts in literature versus non-literature. 3.1 Materials In the second study, we used the same seven corpora and nine texts as in Study 1. 3.2 Results The frequency of all bigrams in each of the seven corpora and nine texts was computed and normalized to account for corpus size. Next, only bigrams that were shared across all materials were selected, resulting in a total of 61 bigrams. The normalized frequencies of these bigrams were compared between the literary and the non-literary texts using a Mann-Whitney test. Those bigrams that yielded significant differences (p < .05) are presented in Table 2. All significant Z-values showed negative scores, with a higher occurrence of these bigrams in literary than in non-literary texts. Next, those bigrams that yielded a significant difference between literary and non-literary texts at p < .01. This selection included 10 bigrams that were submitted
Table 2. Mann-Whitney test between bigram frequencies of literature and non-literature Bigram
U
and in and so and the and what and with as I as it but the by the for a I am I had I will in a in my in the into the
0 10 11 5 3 5 1 6 10 6 5 9 6 6 6 10 2
* p < .05, ** p < .01.
Z −3.33** −2.28* −2.17* −2.81* −3.02** −2.81* −3.23** −2.70* −2.28* −2.70* −2.81* −2.38* −2.70* −2.70* −2.70* −2.28* −3.12**
Bigram
U
Z
is that it to it was me to not the of a of this on a on the out of that I to be to see was a what is with a with the
2 14 10 11 15 4 5 14 14 4 11 12 2 4 11 3 11
−3.12** −1.85* −2.28* −2.17* −1.75* −2.91** −2.81* −1.85* −1.85* −2.91** −2.17* −2.06* −3.12** −2.91** −2.17* −3.02** −2.17*
Max Louwerse, Nick Benesh & Bin Zhang
to a discriminant analysis. Three bigrams classified literary from non-literary text with 100% accuracy. These bigrams and their discriminant function coefficients are presented in Table 3, the classification results in Table 4 and the Z-scores in Table 5.
Table 3. Discriminant function coefficients per bigram Bigram
Discriminant function coefficient
and in and with was a
75.692 78.845 38.229
Table 4. Classification results bigrams
Predicted
Original
Literature Literature Non-literature
100% (9) 0 (0)
Non-literature 0 (0) 100% (7)
Note: Numbers in parentheses are actual counts. Percentage of correctly classified, 100, Eigenvalue, 16.71, Wilk’s λ, .056, χ2, 35.93, p < .001.
Table 5. Discriminant scores per text Text Anna Karenina The Great Gatsby War and Peace Madame Bovary Hamlet Middlemarch Adventures of Huckleberry Finn Letters of Anton Chekhov Lolita Switchboard New York Times Wall Street Journal Santa Barbara Memphis MapTask Edinburgh MapTask TRAINS
Z 4.29 4.26 3.64 3.53 3.49 3.06 2.96 2.95 2.16 −2.44 −2.93 −4.15 −4.21 −5.02 −5.77 −5.83
Chapter 12. Computationally discriminating literary from non-literary texts
The scores per text showed a clear distinction between the literary (positive scores) and non-literary (negative scores) texts. To determine how the results from Study 1 were related to those of Study 2, the stimulus coordinates from the mds results in Study 1 were taken and compared with the discriminant scores of study 2. A strong correlation was found (r(16) = .91, p < .001). This shows that both the lsa and bigram analysis yielded very similar results, despite the fact that lsa emphasized semantic content expressed by lexical items, whereas shared bigrams typically highlight functional items (see also Louwerse, Lewis & Wu, in press). As Crossley and Louwerse (2007) have argued, one of the advantages of a bigram analysis over an lsa analysis is that it allows for getting an insight in the linguistic patterns responsible for differences. The three bigrams variables in the current discriminant analysis suggest that literary texts are typically written in a past tense (“was a”). Moreover, literary texts had a higher frequency of preposed adverbial phrases as indicated by bigrams like “and in”, and “and with”. This may indicate a more frequent use of thematic reorientation, as well as cognitive reorientation, in literary texts compared to non-literary texts (Givón 1993). That conclusion may be premature based on these three variables, but the fact is that a perfect classification was obtained using the frequencies of only these bigrams, very similar to that using the semantic content of the texts. Both Studies 1 and 2 used data mining techniques: on the basis of frequently occurring patterns, texts were classified in literary and non-literary. Can the findings obtained from lsa and bigram techniques be extended to unigrams? And do variables reported in previously published work in literary studies yield a similar classification of the current corpora and texts? These questions are answered in Study 3.
4. Study 3 In the third study, we followed up on a previous one by van Peer (1986). In that, linguistic features in quality literature are compared with those in popular literature, as defined according to their production process and distribution channels. Van Peer (1986) argued that quality literature has a higher number of occurrences of 3rd person narrators, but a lower number of 1st person narrators than popular literature. Moreover, quality literature has a higher number of 1st names and last names than popular literature. Study 3 investigated the frequency of names and narrators in the seven corpora and nine texts. 4.1 Materials Study 3 used the same nine literary texts and seven non-literary corpora as in the previous two studies.
Max Louwerse, Nick Benesh & Bin Zhang
4.2 Results For each text, the frequency of 1st person narrators, 3rd person narrators, first names and family names were computed. First person narrators were operationally defined, following van Peer (1986) using the pronouns “I”, “my”, “mine”, third person narrators by the pronouns “he”, “she”, “his”, “her”, “him”, “hers”. Clearly, this is an operational definition, since pronouns could refer to narrators as well as to characters. First names were identified using 4,275 female first names and 1,219 male first names as well as 88,799 last names from the U.S. Census Bureau data (1994). Name and pronoun frequencies were normalized to account for text size. As in the previous study, a Mann-Whitney test was conducted to determine which of these variables yielded a significant difference between literary and nonliterary texts. The results are presented in Table 6, showing that only pronouns related to the 3rd person narrator yielded a difference. Next, the 3rd person narrator variable was entered in a discriminant analysis. A total of 93.8% of the cases was correctly classified only using the third person pronoun (discriminant coefficient .664). Classification results are presented in Table 7, scores per text in Table 8. The one text that was misclassified is Letters of Anton Chekhov. Adventures of Huckleberry Finn, Hamlet and Lolita were classified correctly, but had similarly
Table 6. Mann-Whitney test Variable
U
First name Last name 1st person narrator 3rd person narrator
17 30 14 0
Z −1.53 − 0.16 −1.85† −3.33**
Note: **p < .01, †p < .07. Negative Z scores denote a higher frequency in literary texts.
Table 7. Classification results third person narrator
Predicted
Original
Literature Non-literature
Literature
Non-literature
88.9 (8) 0 (0)
11.1 (1) 100.0 (7)
Note: Numbers in parentheses are actual counts. Percentage of correctly classified, 93.8, Eigenvalue, 1.97, Wilk’s λ, .337, χ2, 14.695, p < .001.
Chapter 12. Computationally discriminating literary from non-literary texts
Table 8. Discriminant scores per text Text Anna Karenina Madame Bovary War and Peace Middlemarch The Great Gatsby Adventures of Huckleberry Finn Lolita Hamlet Letters of Anton Chekhov New York Times Santa Barbara Corpus Switchboard Wall Street Journal Memphis MapTask Edinburgh MapTask TRAINS
Discriminant score 2.91 2.88 1.73 1.66 1.39 0.42 0.31 −0.09 −0.80 −1.04 −1.16 −1.38 −1.38 −1.77 −1.85 −1.85
low discriminant scores. It is noteworthy that two of these texts (Letters of Anton Chekhov, Lolita) also scored low in the classification in Study 2, and all three of these texts clustered together in Study 1. In fact, a strong correlation was found between the discriminant scores obtained from the texts in Study 2 and 3 (r(16) = .852, p < .001). The main finding of this study is that a near-perfect classification between literary and non-literary texts was obtained and that this classification was similar to that of Study 1 and Study 2.
5. Study 4 Studies 1–3 showed that different computational linguistic techniques (lsa, bi grams, unigrams) obtained very similar results for the 16 texts. Perhaps the similarity can be explained by the texts that were used. Indeed, the argument could be made that Studies 1–3 primarily compared literary texts and spoken dialogue. This argument does not quite hold because newspaper articles were included, but the majority of the materials were dialogues. To account for issues regarding generalizability of the selected materials, a fourth study used the variables from Studies 2 and 3, the three bigrams (“I was”, “and in” “and with”) and the 3rd person narrator, and ran a discriminant analysis on a different set of texts.
Max Louwerse, Nick Benesh & Bin Zhang
5.1 Materials Texts included 119 literary texts, as identified by Zane (2007), without the top 10 literary texts used in the previous three studies. These included A Midsummer’s Night Dream, A Portrait of the Artist as a Young Man, poems of Emily Dickinson, The Ancient Mariner, and The Divine Comedy. Non-literary texts included 55 random texts from Project Gutenberg (Hart 2004) that were not classified by the Library of Congress Classification (lcc) System as Language and Literatures (a-z, with the exception of pn-pz). These included genres identified by lcc as General Works; Philosophy, Psychology, and Religion; Auxiliary Sciences of History; General and Old World History; History of America; Geography, Anthropology, and Recreation; Social Sciences; Political Science; Education, etc., with titles of texts like The Outline of Science, A Text-Book of the History of Painting, Across Unknown South America, Custom and Myth, and History of the United States. As in the previous studies, both the literary and non-literary texts differed in content and genre. 5.2 Results Normalized frequencies were computed for the three bigrams and the third person narrator variables for each of the 174 texts. These frequencies were next entered in a discriminant analysis comparing the literary with the non-literary texts. If the results from the previous studies permit being generalized, these four variables should allow for a higher-than-chance classification into literature and nonliterature. Results again showed that bigrams and third person pronouns allow for a distinction between literary and non-literary texts, with 87.4% of the texts correctly classified. Two of the bigrams and the third person narrator contributed to this classification (Table 9). Classification results are presented in Table 10. These results show that the findings obtained from Studies 1–3 cannot be attributed to specific texts. When a larger and different set of texts is used, bigrams and third person narrators allow for an equally high classification performance of literary versus non-literary texts.
Table 9. Discriminant function coefficients Variables and in and with 3rd person narrator
Function coefficient −.472 .621 .922
Note: Negative scores denote a higher frequency in literary texts, positive scores a higher frequency in non-literary texts.
Chapter 12. Computationally discriminating literary from non-literary texts
Table 10. Classification results bigram and third person narrator 174 texts
Predicted
Original
Literature Non-literature
Literature
Non-literature
89.1% (106) 16.4% (9)
10.9% (13) 83.6% (46)
Note: Numbers in parentheses are actual counts. Percentage of correctly classified, 87.4, Eigenvalue, 1.481, Wilk’s λ, .486, χ2, 154.921, p < .001.
One could argue that in the Studies 1–4 cell sizes are different. For instance, in Study 4, there are far more literary texts than non-literary texts. As stated earlier, discriminant analysis is relatively insensitive to groups not being of equal size. To illustrate this, the 55 non-literary and a random sample of 55 literary texts were entered in the discriminant analysis. Classification performance was the same, with 85.5% of the cases correctly classified. To extend the generalizability of the findings reported so far, a final analysis was conducted in Study 5. 6. Study 5 Study 1 showed that lsa allows for a clustering of the corpora and texts into literary and non-literary, Studies 2 and 3 showed that the same classification can be obtained using shared bigrams and even unigrams. Study 4 extended these findings by considering a different set of literary and non-literary texts, thereby ruling out the possibility that the results obtained in Studies 1–3 are due to genre (dialogue and newspaper versus novels) rather than literary versus non-literary differences. However, the question can be raised whether these studies compare literature with non-literature. Perhaps the classification results can be explained by a comparison of narrative (literary texts) and non-narratives (dialogues, newspaper articles, histories, textbooks). To answer these questions Study 5 further investigated the literary aspect of the texts by comparing two groups of narratives, what van Peer’s (1986) called ‘quality literature’ and ‘popular literature’. 6.1 Materials The same 119 quality literature texts from Study 4 were used. As a popular literature comparison, 42 Star Wars novels (1979–2003) were made electronically available. These were part of a collection of novels approved by George Lucas and were contiguous with the original three Star Wars movies. Much like a long
Max Louwerse, Nick Benesh & Bin Zhang
running Star Wars TV series with new episodes taking into account previous ones, the novels revolve around the main characters’ adventures to save the galaxy from various catastrophes. The authors of the novels vary in many dimensions, such as notoriety, influence on the overall Star Wars time line, and general experience in the science fiction genre. 6.2 Results Of the four variables (three bigrams and third person narrator) entered in the discriminant analysis only the bigram “and in” was kept in the model (discriminant function coefficient = 50.517). This variable alone classified 86.3% of the cases correctly. Classification results are given in Table 11. Five out of the 119 literary texts were misclassified on the basis of the frequency of the bigram “and in”. These were The Importance of Being Earnest, Uncle Vanya, To Kill a Mockingbird, poems of Emily Dickinson and Hedda Gabbler. The reason for this misclassification is not clear. Though three of these five texts are not novels but plays and poems, the remaining 114 literary texts contain many plays and poems that were classified correctly, such as King Lear, Henry IV, Henry V, The Aeneid, The Iliad and The Ancient Mariner. Study 5 showed that when distinguishing quality from popular literature, a high classification performance can be obtained using the normalized frequency of one single bigram (86.3%). Moreover, the classification results show that the quality literary texts are unlikely to be classified as popular literature (95.8%), but popular literature may get classified as quality literature (40.5%). That is, what is quality literature remains quality literature, but in some cases what is popular literature could be considered quality literature. 7. Conclusion According to Fowler (1996: 21) “Literature is a creative use of language”. The five studies presented in this paper show that different computational linguistic Table 11. Classification results quality literature and popular literature
Predicted
Original
Literature Star Wars
Literature
Star Wars
95.8 (114) 40.5 (17)
4.2 (5) 59.5 (25)
Note: Numbers in parentheses are actual counts. Percentage of correctly classified, 86.3, Eigenvalue, .448, Wilk’s λ, .691, χ2, 58.677, p < .001.
Chapter 12. Computationally discriminating literary from non-literary texts
techniques allow for a high performance classification into literary and non-literary texts on the basis of the language in literary and non-literary texts. Study 1 showed that higher-order occurrence techniques like lsa can cluster corpora and texts, Study 2 showed similar results can be obtained using three shared bigrams, and Study 3 showed even pronoun frequencies allow for an accurate classification. Moreover, despite the different techniques used in Study 2 and 3, a correlation of discriminant scores was obtained. Study 4 showed that this classification performance cannot be attributed to the selection of texts. On the contrary, when a larger set of corpora and texts of literary and non-literary variety were used, classification performance remains high. In fact, even when two sets of narratives (quality literature and popular literature) were compared, classification performance remains considerably higher than chance (86.3%). Moreover, what Study 5 showed is that quality literature was unlikely to be classified as popular literature, but what is popular literature might be classified as quality literature. The finding that different computational linguistic techniques yield similar results is important for future analyses. For instance, both a focus on lexical items (content words) and a focus on functional items (pronouns, conjunctions) yielded similar results (see also Louwerse, Lewis & Wu, in press). An overview of some of the differences is presented in Table 12. The five studies reported in this paper served a number of purposes. First, they presented some straightforward computational linguistic techniques like Latent Semantic Analysis, shared bigram frequencies and word frequencies available to researchers investigating linguistic patterns in text and discourse. Secondly, the studies reported here shed light on the question whether literary texts can be distinguished from non-literary texts using language features, not in a selected text sample but throughout the text. Twenty years ago, if somebody asked in his office on the second floor immediately left from the squeaky stairs what constituted literature, the reply would have been with terms like “aesthetic”, “deconstruction”, “fictionality”, “defamiliarization”, “syuzhet”, “foregrounding” and “literariness”. Twenty years later, we can reply with the bigram “and in”, an empirically valid answer. Whether Table 12. Differences between computational linguistic techniques Sensitivity to paragraph size Sensitivity to text size Variables user defined Sensitivity to exact string match Sensitivity to lexical items Sensitivity to grammatical items Sensitivity to linguistic patterns
Word count
Shared bigrams
LSA
– – + + + + –
– – – + – + +
+ – – – + – –
Max Louwerse, Nick Benesh & Bin Zhang
that computational linguistic answer is utterly naïve, as van Peer (1989) suggested, we leave open for discussion.
8. Acknowledgments This research was supported by grant NSF-IIS-0416128. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funding institution.
References Allen, J. & Heeman, P.A. 1995. TRAINS Spoken Dialog Corpus [CD-ROM]. Philadelphia PA: Linguistic Data Consortium. Crossley, S.A. & Louwerse, M.M. 2007. Multi-dimensional register classification using collocations. International Journal of Corpus Linguistics 12: 453–478. Du Bois, J.W., Chafe, W.L., Meyer, C. & Thompson, S.A. 2000. Santa Barbara Corpus of Spoken American English [CD-ROM]. Philadelphia PA: Linguistic Data Consortium. Fabb, N. 2002. Language and Literary Structure: The Linguistic Analysis of Form in Verse and Narrative. Cambridge: CUP. Fowler, R. 1996. Linguistic Criticism. Oxford: OUP. Givón, T. 1993. English Grammar: A Function-based Approach. Amsterdam: John Benjamins. Godfrey, J.J. & Holliman, E. 1993. Switchboard-1 [CD-ROM]. Philadelphia PA: Linguistic Data Consortium. Graff, D. 1996. North American News Text Corpus [CD-ROM]. Philadelphia PA: Linguistic Data Consortium. Green, M. 1987. Tolstoy and Nabokov. The morality of Lolita. In Nabokov’s Lolita, H. Bloom (ed.), 13–33. New York NY: Chelsea House. Hart, M. 2004. Project Gutenberg. Retrieved November 20, 2007, from http://www.gutenberg. org/ Human Communication Research Centre. 1993. HCRC Map Task Corpus [CD-ROM]. Philadelphia PA: Linguistic Data Consortium. Jefferson, A. & Robey, D. 1986. Modern Literary Theory: A Comparative Introduction. London: Batsford. Jurafsky, D. & Martin, J.H. 2000. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition. Upper Saddle River NJ: Prentice-Hall. Kintsch, W. 2002. On the notions of theme and topic in psychological process models of text comprehension. In Thematics: Interdisciplinary Studies, M.M. Louwerse & W. van Peer (eds), 157–170. Amsterdam: John Benjamins. Landauer, T.K. & Dumais, S.T. 1997. A solution to Plato’s problem: The latent semantic analysis theory of the acquisition, induction, and representation of knowledge. Psychological Review 104: 211–240.
Chapter 12. Computationally discriminating literary from non-literary texts
Landauer, T.K., McNamara, D.S., Dennis, S. & Kintsch, W. (eds). 2007. Handbook of Latent Semantic Analysis. Mahwah NJ: Lawrence Erlbaum. Landauer, T.K. 2002. On the computational basis of learning & cognition: Arguments from LSA. The Psychology of Learning & Motivation 41: 43–84. Louw, B. 1993. Irony in the text or insincerity in the writer? In Text and Technology. In honour of John Sinclair, M. Baker, G. Francis & E. Tognini-Bonelli (eds), 157–176. Amsterdam: John Benjamins. Louwerse, M.M., Bard, E., Steedman. M. & Graesser, A.C. 2004. Memphis Multimodal MapTask Corpus [DVD]. Memphis TN: University of Memphis. Louwerse, M.M., Lewis, G. & Wu, J. In press. Unigrams, bigrams and LSA: Corpus linguistic explorations of genres in Shakespeare’s plays. In New Directions in Literary Studies, W. van Peer & J. Auracher (eds). Cambridge: Cambridge Scholars Publishing. Louwerse, M.M. & Van Peer, W. 2006. Waar het over gaat in cijfers. Kwantitatieve benaderuingen in tekst- en literatuurwetenschap. (What it is about in numbers: Quantitative approaches in text- and literary studies). Tijdschrift voor Nederlandse Taal- en Letterkunde 122: 21–35. Louwerse, M.M. & Van Peer, W. In press. How cognitive is cognitive poetics? The interaction between symbolic and embodied cognition. In Cognitive Poetics, G. Brone & J. Vandaele (eds). Berlin: Mouton de Gruyter. Louwerse, M.M. 2004. Semantic variation in idiolect and sociolect: Corpus linguistic evidence from literary texts. Computers and the Humanities 38: 207–221. Manning, C.D. & Schutze, H. 1999. Foundations of Statistical Natural Language Processing. Cambridge MA: The MIT Press. Rowe, J.A. 1988. Equivocal Endings in Classic American Novels: The Scarlet Letter; Adventures of Huckleberry Finn; The Ambassadors; The Great Gatsby. Cambridge: CUP. Sinclair, J. 1966. How to take a poem to pieces. In Essays on Style and Language, R. Fowler (ed.), 68–81. London: Routledge and Kegan Paul. Sinclair, J. 2004. Trust the Text. London: Routledge. Stubbs, M. 2005. Conrad in the computer: Examples of quantitative stylistic methods. Language and Literature 14(1): 5–24. U.S. Census Bureau. 1994. 1990 Census Name Files. Retrieved November 30, 2007, from http:// www.census.gov/genealogy/names/names_files.html Van Peer, W. 1986. Pulp and purpose. Stylistic analysis as an aid to a theory of texts. Linguistics and the study of literature. In Linguistic Contributions to the Study of Literature, T. D’Haen (ed.), 268–286. Amsterdam: Rodopi. Van Peer, W. 1989. Quantitative studies of style: A critique and an outlook. Computers and the Humanities 23: 301–307. Zane, J.P. (ed.). 2007. The Top Ten: Writers Pick Their Favorite Books. New York NY: W.W. Norton.
chapter 13
Metaphors and software-assisted cognitive stylistics Michael Kimmel This paper reports on an ongoing project on literary metaphor that applies analytical tools from cognitive linguistics (cl) to six English novellas and presents a case study. The project focuses on complex tropes, metaphors, coherence and cohesion in text units as well as text cues that involve the reader at a bodily level. Its hallmark is systematic and multi-level metaphor tagging and analysis with the qualitative coding software ATLAS.ti 5.2. My aim here is to raise three methodological points of increasing specificity. First, I explain the general benefits of a software-assisted qualitative analysis. Second, I present a flexible and powerful approach for metaphor coding (which is extendable to non-literary applications). Third, I illustrate the gains from applying the approach to literature at a descriptive level, then in a more integrative literary analysis, and finally with regard to a novella’s general trends in metaphor use. Keywords: cognitive linguistics, cognitive stylistics, conceptual metaphor, image schemata, software-assisted analysis, authorial style.
1. Literary metaphor from a cognitive linguistic (cl) angle Metaphors, in literature as in everyday speech, depend upon a word or phrase (a source- or vehicle-term) creating semantic tension with its cotext or context. Successfully resolving this tension is a unique faculty of the human mind and requires properties from the vehicle’s conceptual domain to be mapped to an explicit or implicit target domain. This invariably selective feature transfer may include attributes, images, inferential structure, or affect structure. cl demonstrates that thousands of everyday metaphorical expressions constitute highly systematic sets (e.g., life is a journey, life is a day, life is a plant; good is up, good is bright), each with a single underlying logic. This systematicity points to production/reception schemata for
Michael Kimmel
metaphors called conceptual metaphors. According to Lakoff and Turner (1989), most literary metaphors are rooted in everyday conceptual metaphors and not based on “strangification” vis-à-vis everyday cognition. The prototypical cl analysis of metaphor gathers expressions from across a text or corpus and then draws together patterns similar either in imagery or inferential entailments. About two dozen case studies have applied this technique to literary narrative and its recurrent meaning patterns (see Steen & Gibbs 2004). Traditionally-minded literary theorists find this approach wanting (e.g., Downes 1993), partly because they emphasize close stylistic readings of what specific metaphors do in context, and partly because they believe that old-fashioned methods receive only a cognitive re-labeling. My position in this debate is that cl theory can genuinely enrich a stylistic perspective by paying attention to systematicity at a deeper conceptual level as well as by connecting with experimental research, but that discourse methods of metaphor analysis are needed to give full contextualization to cl. Presently, several shortcomings in cl-based literary research need to be remedied: 1. It is typical of case studies on literary metaphor for scholars to select metaphor patterns that happen to strike them and let others go unnoticed (e.g., Stockwell 2002), both because some seem too banal to be of literary relevance and because their annotation technique is unsystematic. 2. Most analysts pick out a single level of metaphorical cognition in a case study, often only recurrent image schemata like path (e.g., Freeman 1993). Yet, metaphor is a multifaceted phenomenon inviting a complex analysis at parallel levels (notably imagery patterns and inferences). 3. Pulling similar metaphors together from anywhere across a text is a legitimate analytical technique. Yet, further attention should be paid to an exploration of textual cohesion effects, both in the sense that different metaphors in a text segment may blend into more than the sum of their parts (Danaher 2003) and that similar metaphors might acquire a special potency when adjacent. 4. Although metaphor has been analyzed as a theme-cueing device, its role for creating abstract narrative dynamics and basic event structure has not much been in focus (Kimmel 2005). 5. Previous studies neglect what one may call an author’s “metaphorizing strategy”, i.e., quantifiable general trends regarding how a text or genre uses metaphor (Do they tend to be synesthesias, comparisons, part of multi-metaphor scenarios? Is there a wide variety of metaphors? Are there singularly frequent patterns? Are patterns equally distributed? Do they cluster?, etc.).
Chapter 13. Metaphors and software-assisted cognitive stylistics
State-of-the-art tools should thus be designed to explore literary metaphor applying discourse-related techniques, using a multi-level code scheme, and with full systematicity in capturing a work’s metaphors, but without resorting to full automation and relinquishing scholarly intuition about context.
2. Methodology 2.1 Qualitative analysis Using software-assisted qualitative analysis means carefully reading a text and applying codes to it that identify particular phenomena (or using the text inductively to develop a set of codes). Codes are applied with context-sensitivity, although auxiliary searches for predefined word-lists, as in corpus linguistics, are also possible. The codes allow the later retrieval of quotation lists, easy browsing of quotation context, and complex pattern searches for co-occurring codes (by exact Boolean logic, for codes that share a predefined code family or simply for codes that overlap or are close in the text). The search tool thus enables complex theoretical queries, e.g., “which metaphor types/functions dominate in dialogues (as opposed to descriptions)?” 2.2 Metaphor coding in atlas.ti 5.2 Using a recent definition of metaphor (Pragglejazz Group 2007), I coded all words or phrases that have another more (spatio-physical, sensorial, etc.) basic reference than the expression’s contextual meaning. For this, in atlas.ti, a text segment is marked (left) and a code from the list (right) dragged onto it:
Figure 1. Screenshot of the coding interface.
The metaphor units were a clause or at most a sentence. Whenever a sentence involved several independent metaphors, these were tagged separately. To deal with the fact that there are degrees of metaphoricity, a category of borderline phenomena was specially marked, including phrasal verbs and cases of polysemy.
Michael Kimmel
2.3 Coding of metaphor types First, text units identified as metaphor were coded by general metaphor type. The basic types were “conceptually productive metaphor” and “metaphor with synesthetic properties”, in contradistinction to one-shot “idioms” that engender no similar expressions, and thus ramain conceptually unproductive. Linguistic surface properties like “comparison” and cognitive properties like “complex trope” (e.g., irony between two metaphors) or “metaphor scenario” (a type of compound image) were coded as well. All general codes were created for easy quantitative surveying of the kinds of metaphoric phenomena that dominate a text. 2.4 Coding of metaphor details Next, two specific codes were assigned to each metaphor, the source and target domains. This is based on the assumption by Lakoff and Johnson (1999) that to identify a metaphor one needs to reconstruct its source, its target, and then what gets mapped between them. The specific codes aim at reconstructing the specific mapping in the main qualitative analysis. To illustrate type and detail coding, the expression “this conversation is leading nowhere” receives the general code “conceptually active metaphor” and then detail codes “source: paths” and “target: conversation”, which in turn point to the formula conversations are paths. Formulas like this are the fundament for describing and retrieving the most frequent mappings in a novel. However, coding itself remains compositional, as the code system features no metaphor formulas. The latter are reconstructed only later by checking all frequent co-occurrences between sources and targets. This research strategy is essential for any complex study of literary metaphor. A full-scale coding of a novel may yield hundreds of conceptual patterns and the alternative “one metaphor type-one code” strategy would make the code list explode beyond manageability. 2.5 Multi-tier coding of source domains As further innovation, source domain codes were applied at two conceptual tiers, which reflect two cognitive “layers” of metaphor processing. For instance, the “state ship confronted an iceberg” invites both path and collision image schemata (shared with non-nautical metaphors such as “running into a wall of silence”) and wider cultural knowledge about the exemplar ship navigation, crews, and captains (shared with any ship metaphor, independent of collision). The one layer is the image-schematic core representation that “carries” the basic ontological structure of a mapping, and the other layer, the cultural exemplar that piggybacks on it by adding richer knowledge and inferential entailments. Coding metaphors at both levels brings out multiple similarities between metaphors.
Chapter 13. Metaphors and software-assisted cognitive stylistics
3. Case study: Mapping patterns in qualitative perspective 3.1 Plot synopsis of one novella The Turn of the Screw, one of Henry James’ best-known works, is a psychological ghost story situated in a mid 19th-century “Gothic” estate where a young governess looks after ten-year old Miles and his younger sister, Flora. The children are described as precocious, sweet, talented, and enthralling. Although a framing tale is present – an old gentleman recounts the adventure decades later – the story itself is told from the perspective of the governess and focuses much on her inner feelings, crises, as well as her resolve. Soon, ghost apparitions of the “fallen” Ms. Jessel and her seducer, the house teacher Quint, begin challenging the governess, although no one else admits to seeing them. The governess becomes convinced of a secret collusion between the children and their ghostly former caretakers, which she almost obsessively tries to uncover. She begins a struggle to save the children from the immoral influence, aided by the housekeeper Mrs. Grose, carefully keeping her countenance, but scanning desperately for corroborative signs. As the governess’ mounting turmoil leads her to confront Flora with an explicit question, the girl reacts hysterically and is taken to town for recovery. In the story’s culminating scene, the governess pressures Miles to confess Quint’s presence, upon which Miles breaks down and dies of heart attack (whether of possession by the ghosts, of fear, or by smothering in an embrace remains open). 3.2 Getting a quick overview of metaphor-based themes Many narrative key themes are metaphorically reflected in the novella. For a useful quick-scan, all target domains may be compacted into wider thematic groups and their frequency counted. Overviews such as this one are helpful to get a grip on the data, discover main story themes (= targets), and to structure the write-up by supplying main headers to group metaphorical mappings by. As to the most frequent targets, the many communication metaphors appear in dialogues, but also describe the governess’ manipulation of Mrs. Grose as well as her feeling manipulated by the children. The strong introspective focus accounts for why the story is extremely dense in emotion and cognition metaphors. Finally, the knowledge-related metaphors result directly from the theme of the governess’ quest for knowledge. Note that a more differentiated listing would just as easily be possible. It would show that the themes reflected metaphorically include incertitude of knowledge, obsession, complex psychodynamics and emotional dilemmas, childhood innocence and over-protectivenesss, as well as fear of immorality and sexuality.
Michael Kimmel
Table 1. Theme overview Target domain group
hits
Interaction and communication Psychological states and dynamics Knowing or not Emotions Characteristics of persons Action goals (such as discovering the secret) Event descriptions more generally External appearance of persons and things Other cognition-related Morality, abjectness, seduction
202 165 160 154 104 87 67 45 45 28
3.3 Procedure of the qualitative analysis The actual qualitative groundwork of the analysis uses metaphor detail codes to reconstruct all relevant mapping patterns of a text. By the logic of compositional coding, metaphorical mappings can be identified only via co-occurrences between metaphor sources and targets that each give rise to one kind of mapping. For this, a list of source-target combinations is printed from atlas. ti. For example, the target domain “knowledge” can be listed and sorted by all the source domains it overlaps with (the co-occurrence frequencies are given in brackets). t: Knowledge [117] IMAGE-SCHEMATIC SOURCE DOMAINS RICH SOURCE DOMAINS si: Objects, entities and substances [25] sr: Picture si: Paths, pace, carrying on paths [25] sr: Possession si: Containment, engulfment, breach [22] SYNESTHETIC SOURCE DOMAINS si: Forces and momentum [21] ss: Seeing / vision si: Up-down [9] ss: Bright-dim/color si: Surface [5] ss: Grasp / touch si: Together-apart [5] ss: Heavy-light
[6] [6] [30] [21] [9] [5]
Figure 2. List of code co-occurrences
Here a pre-selection filter of 5 hits per combination is applied to limit what enters into the analysis. Text quotations for each mapping from the list are checked to see whether the codes apply similarly to each context. Then, for each mapping, quotations are compiled and sorted by sub-variants, e.g., different sub-scenarios
Chapter 13. Metaphors and software-assisted cognitive stylistics
of a core mapping or different sets of concepts that get mapped from a given source domain. Next, coherent sets of expressions are subsumed under a metaphor formula at the appropriate level of abstraction, e.g., knowing is grasping, knowledge is bright or ideas are objects. Finally, cognitive and discursive functions occurring within each formula are worked out, e.g., highlightinghiding, information compression, inference, grabbing attention, creating humor, or emotionalizing. 3.4 Grouping and analysis of selected mappings For further illustration, I selected a few mappings contributing to a main story theme in The Turn of the Screw and will now list them by target. Each of the targets such as “children”, “interaction”, etc. shares a high-level schema (underlined) that spawns various mappings (in small caps). The actual metaphorical words appear in boldface in each exemplary phrase: a. The children are dear creatures the children are angels → “one of Raphael’s holy infants, who looked from one of us to the other with placid heavenly eyes” (paragraph 128) the children are bright creatures → “radiant image of my little girl” (paragraph 128) the children’s behavior is sweet → “I shall never forget the sweetness and gaiety with which he brought out the word” (paragraph 810) b. Interaction between children and the governess enthrallment by the children is force interaction → “You WILL be carried away by the little gentleman!”. (paragraph 136) enthrallment by the children is being possessed → “that he must know how he really, as they say, ’had’ me”. (paragraph 802) enthrallment by the children is a magic spell → “with his spell all scattered” (paragraph 764) c. Interaction between ghosts and children captivation of the children’s souls by the ghost is possession → “but he has lost you forever!” (paragraph 1659)
. As I use a digital version of The turn of the Screw to carry out the analysis, only paragraph numbers are available.
Michael Kimmel
d. Interaction between the governess and children interaction and communication between governess/children is violence → “The wretched child had spoken exactly as if she had got from some (paragraph 1327) outside source each of her stabbing little words” interaction and communication between governess/children is war → “at the drop of my victory and all the return of my battle” (paragraph 1647) In these examples metaphors contribute directly to the story plot by constituting a main theme. However, many other metaphors may only serve basic discourse functions. The fact that everyday language is inescapably woven through with metaphors makes it important to distinguish those used with more specific purposes. In addition, conceptual metaphors appear both in fully conventional phrases and in elaborated (= “literary”) form. Based on the example communication is object transfer (on a path) Table 2 below illustrates the resulting four metaphor types: Table 2. Analytical types of metaphor communication is object transfer
Basic function (no key theme implied)
Conventional use of conceptual metaphor
(1) “They go into no (2) “But nonetheless, between particulars.” (paragraph 177) Miles and me, it’s now all out.” (paragraph 1033) (theme: enigma)
Elaborated conceptual metaphor
(3) “If he had been wicked (4) “She offered her mind to he would have “caught” it, my disclosures as, had I wished and I should have caught to mix a witch’s broth and it by the rebound” proposed it with assurance, (paragraph 297) she would have held out a large clean saucepan.” (paragraph 800) (theme: manipulation)
Key theme-related
The table differentiates for qualitative analysis (1) perfectly conventional uses of conceptual metaphor, (2) creatively elaborated basic conceptual metaphors, (3) conventional theme-related metaphors, and (4) creatively elaborated and themerelated metaphors. After description, mapping patterns invariably need to be interpreted in their wider context. For example, the “brightness” and “angelic” nature of the children becomes increasingly ambivalent as the governess becomes suspicious of their conspiracy. In a contextualized view, the pattern actually creates an effect of tension.
Chapter 13. Metaphors and software-assisted cognitive stylistics
4. Integrative analysis of metaphor patterns Once the basic descriptive analysis of the most frequent metaphors is complete and the researcher is closely familiar with the data, cl theory offers several ways of advanced data synthesis. The researcher first explores the data, builds hypotheses and tests them by using search queries, co-occurrence lists, word searches, full-text navigation, as well as customized data filtering and code grouping in atlas.ti. 4.1 A nalysis based on affinities between image schemata (= source domains) One popular kind of analytical technique identifies key image schemata of a work (Freeman 1993, 1995), based on the assumption that the different metaphorical mappings, of which they are source domains, evince a similar logic and jointly contribute to a single narrative theme or symbolic node. For example, I found the up-down image schema striking enough to single it out for analysis. It furnishes the source domain of metaphors describing the governess’ oscillation between elation and despondency, and the moral “fall” of Quint and Jessel who want to pull the children “down”. In addition, symbolically imbued scene descriptions of high towers (i.e., not strictly metaphors) and the like are frequent. A possible connection between the up-down schemata is the irony that Quint, the fallen man, appears high up, in power. More generally, as image schemata are usually value imbued, the up-down node creates a permanent textual oscillation between positive and negative associations. This becomes a major resource for James’ master trope of ambiguity and underwrites the hypothesis that up-down functions as a kind of symbolic node. 4.2 Analysis based on affinities between full conceptual metaphors Affinities between conceptual metaphors are equally interesting. In some literary works, even a portion of plot-driving story macrostructure can be deduced from these (Kimmel 2005), particularly since force-, path-, or container-based mappings often co-specify movements between narrative spaces and the like. In the Turn of the Screw, it is force-dynamic conceptual metaphors that create a complex schema, which sheds light on the governess’ character and her constant psychological dilemma. The metaphors consistently appear in her introspective passages, . It is thanks to the compositional coding that an independent search of image-schematic source domains is possible.
Michael Kimmel
albeit in constantly shifting patterns, as she oscillates between reluctance and fear, despondency, fighting, and the urge to shield the children. These conceptual metaphors are illustrated below: desires are impulsive forces → “the strange impulse that I lately spoke of as my temptation” (paragraph 791) psychological pressure is a force → “assailed with apprehensions” (paragraph 1479) controlling emotions is vying with a counterforce → “the duty of (paragraph 693) resistance to extravagant fancies” allowing emotional release is letting go → “were I to let myself go even now” (paragraph 760) venting or showing emotions is letting them out → “my lamentation (paragraph 688) overflowed” being subject to an emotion is being locked in it → “Agitation […] (paragraph 291) certainly had held me and driven me”
A basic affinity between the force metaphors for emotions is obvious, which jointly describe an inner struggle (“letting go emotionally”). A more advanced kind of qualitative analysis can reconstruct the inferential connections that constitute parts of an implied “action chain”. The governess is subject to forceful impressions; this produces emotional impulses; she is striving to control or escape these, but then ultimately needs a periodical outlet (the subtext being that her own strong ideas and desires create the conflict). The dynamics of the governess’ inner world is made of a series of coherent image schemata that create a sort of oscillating mini-clip before the mind’s eye, going through an “enclosing mode” to an “outbreak mode” of emotion. As with the up-down schemata, the emotionrelated force metaphors seem to be a major cue for creating a cyclical structure and constant narrative “disequilibrium”. These and other cues jointly constitute a major basis for the narrative’s mounting suspense, resulting from the sustained back-and-forth motion. They are part of what drives the plot forward. 4.3 Heuristic discovery of macro-zones where metaphors cluster Before entering into an interpretive analysis, one may be interested in identifying distributions and focal zones of metaphoric activity. For an overview, one may use cumulative metaphor counts (plotted via an spss for Windows data export) and mark the steep parts of the curve representing the highest metaphor density: The key macro-zones (in rectangles, which correspond to several thousands of characters) are then re-entered in the text for qualitative exploration or the graph is used to compare two or more texts with regard to the metaphor distributions.
Chapter 13. Metaphors and software-assisted cognitive stylistics
1.200
Cumulative sum of metaphors
1.000 800 600 400 200 0
0
5.000.000 10.000.000 15.000.000 Begin position of metaphor
20.000.000
Figure 3. Cumulative sum of metaphors and focal zones.
4.4 Cohesion-based exploration I: Micro-zones of metaphor clustering A prime candidate for focal metaphoric micro-zones are segments involving textually cohesive and conceptually coherent metaphors. This means adding a textual adjacency constraint to what is discussed in 4.2. Although a general search for ontologically coherent and adjacent metaphors is possible, let me illustrate the principle by a hypothesis-based search. The query begins by choosing a suspected cluster type like “all clusters involving the source paths and the target action goals”, defining an adjacency window size considered acceptable, and by searching for all passages with, say, three or four cohesive metaphors. This results in so-called metaphor scenarios in which independently meaningful metaphors co-specify and mutually enrich each other. “It was as if, at moments, we were perpetually coming into sight of subjects before which we must stop short, turning suddenly out of alleys that we perceived to be blind, closing with a little bang that made us look at each other – for, like all bangs, it was something louder than we had intended – the doors we had indiscreetly opened. All roads lead to Rome, and there were times when it might have struck us that almost every branch of study or subject of conversation skirted (paragraph 898) forbidden ground.” . Borderline cases appear in italics.
Michael Kimmel
Locating such passages prepares for a qualitative analysis. In the present example we may look at how path and container logic is elaborated into a complex image and how a complex inferential matrix is created that characterizes how the governess and the children communicate. 4.5 Cohesion-based exploration II: Image schema mesh A search query can also be based on “image schema mesh”, i.e., the assumption that image-schematic source domains like force or path, when they cluster in a single passage, are particularly likely to impact the cognitive model of the reader. This is independent of specific metaphor targets, as in 4.1. One may, as before, run a general search for meshing sources of any kind, so as to determine their frequency or types, or run a hypothesis-based search. To illustrate the theoretical import of image schema mesh, two hypotheses generated from my coding were that (i) the Turn of the Screw frequently uses the stylistic strategy of creating metaphor fields with several synesthesic metaphors and that (ii) these in turn are likely to involve cross-modal combinations of sensory imagery. To identify cross-modal metaphors of all sorts of the image schema type called axial, I grouped 20 codes referring to synesthetic source domains (bright-dim, hot-cold, bitter-sweet, loudsilent, sharp-blunt, etc.), which resulted in instances like the following: “It was a crisp, clear day, the first of its order for some time; the night had brought a touch of frost, and the autumn air, bright and sharp, made the church bells (paragraph 909) almost gay.”
Only four such passages were found, which refutes my frequency-related intuition. Their detailed analysis revealed that my other intuition regarding cross-modality was also only half supported, as two of the synesthetic clusters stay within the visual modality, while only two others connect tactile and visual, as in the quote.
5. Literary metaphorizing strategy So far, I have focused on the content level of metaphor and complex meanings obtainable from it. I now turn to the general profile of literary metaphor indicating how an author uses metaphor stylistically at a more abstract level at which texts can be fruitfully compared. A set of quantifiable parameters is arrived at by counting the metaphor type codes created for this purpose. Although researchers quickly see that the Turn of the Screw is strikingly rich in metaphor, they may want to get a more systematic picture, because intuitive impressions are insufficient for comparing texts or genres. To make the comparison-related benefits of the method tangible, I compare the text with
Chapter 13. Metaphors and software-assisted cognitive stylistics
Herman Melville’s Billy Budd, a novella similar in metaphor use, and Ian McEwan’s Cement Garden, which is strikingly different in most parameters. 5.1 Metaphor density and diversity of mappings Two key indicators are metaphors per word (density) and the number of metaphor mappings from the qualitative analysis (diversity). Table 4. Basic frequency and diversity data
METAPHOR UNIT FREQUENCY
Text words Turn of the Screw Billy Budd Cement Garden
42,899 30,638 42,697
Metaphor units
Metaphors per word
1,021 788 385
0.024 0.026 0.009
MAPPING DIVERSITY Metaphor formulas 54 68 17
The left boldfaced column indicates that metaphor codes per text word are strikingly similar between James and Melville, pointing to a similar reliance on metaphor and a stylistic strategy using it as major resource. McEwan’ s use of metaphors is almost only a third of Melville’ s. The right boldfaced column indicates that Melville exhibits higher diversity (and, with a lower metaphor total, fewer hits per mapping) than James’. McEwan resorts to a strategy with an extremely low diversity of recurrent mappings (in his case due to a high percentage of mappings with 1–3 hits only), therefore hardly relying on metaphor to cue story themes and using it more for ad hoc purposes. 5.2 Surface language: Comparison cl treats comparisons and metaphors not as fundamentally different, but as potential ways to express conceptual metaphors, which mainly differ at the linguistic surface. Although the two are related, the percentage of comparisons still remains an interesting indicator of authorial style. . All kinds of mappings are counted, including comparisons, synesthesia, and image mappings. . Mappings with more that four hits were counted for Turn of the Screw and Billy Budd, for The Cement Garden all above two hits because of the anyhow lesser number. . One may of course further interpret these findings: In Turn of the Screw, metaphor density, for example, follows from the novella’s psychological aspect (introspection and character description typically draw on metaphorical concepts), the complexity of its language, and the earlier illustrated fact that several themes rely on metaphor more or less exclusively.
Michael Kimmel
Table 5 indicates that McEwan rates by far highest in comparisons in relation to all metaphors, making them quite distinctive of his style. Table 5. Comparison and metaphor in the narrow sense
Metaphor units
Comparisons
Percentage of comparisons
1,021 788 385
83 104 86
8% 13% 22%
Turn of the Screw Billy Budd Cement Garden
5.3 Metaphoric image mappings and creativity Literary metaphor is said to go beyond everyday metaphor in the way it creates its aesthetic appeal. One key measure for this is the incidence of image metaphors, i.e., specific “one-shot” mappings with rich structure that is not simply imageschematic. Here are two examples from James: “it was only the relief that a snap brings to a strain or the burst of a thunderstorm (paragraph 904) to a day of suffocation.” “but a small shifty spot on the wrong side of it all still sometimes brushed my (paragraph 697) brow like the wing of a bat”
These image metaphors map rich experiential structure of a quite unique sort and are in no way conventional. A quantitative comparison of image mappings in the three novellas results in striking differences: Table 6 indicates that image mappings are rather infrequent in James’s work, and somewhat higher in Melville’s. On the other hand, the disproportionally high ratio of metaphors rich in detailed imagery in McEwan’s is most striking. High metaphoric creativity may be evident in some image mappings, though others simply create concreteness (e.g., “he was lead away like a blind man”). Table 6. Image mappings Turn of the Screw Billy Budd Cement Garden
Metaphor units
Image mappings
Ratio of image mappings
1,021 788 385
15 39 72
1.47% 4.95% 18.7%
Creativity per se is best indexed by metaphor elaboration, extension, negation, and composing (Lakoff & Turner 1989: 67–72). However, examples as the following from James require a qualitative case-by-case analysis to understand how the conventional is transformed:
Chapter 13. Metaphors and software-assisted cognitive stylistics
“It was the idea, the second movement, that led me straight out, as I may say, of (paragraph 477) the inner chamber of my dread.” “that brought back to me, long enough to catch it, the feeling […] She offered her mind to my disclosures as, she would have held out a large clean saucepan.” (paragraph 800) “I find that I really hang back; but I must take my plunge.” (paragraph 764)
The first two creative elaborations are based on conventional mappings (feelings are contained inside, communication is object transfer), but their exemplar images are quite creative and striking. In the third example, two conventional metaphors are creatively combined based on an inherent affinity (storytelling is a journey, courageous action is a dive). Future case studies should aim at quantifying such elaboration patterns. 5.4 Literary complexity I: Metaphor cluster incidence The metaphor clusters discussed before in qualitative perspective also offer meaningful quantitative indicators. Asking how frequent metaphor clusters are and how large they get is informative to the extent that not all texts favor complex metaphor interaction by making them cohere. Table 7. Metaphor clusters Metaphor Metaphors units in clusters Turn of the Screw Billy Budd Cement Garden
Percentage of metaphors in clusters
1,021 499 48.9% 788 304 38.2% 385 26 6.8%
Largest cluster size sevens sevens doubles
This table shows that in the 19th-century novellas the high rate of metaphor clusters, related to the high metaphor density, suggests that the stylistic device of clustering needs to be explored qualitatively, whereas in the Cement Garden it is evident that this will provide little further insight about trends typical for the text. 5.5 Literary complexity II: Metaphor scenario incidence Metaphor scenarios are metaphor clusters with the added constraint that at least two adjacent units are complementary (and often give rise to a single complex image).
Michael Kimmel
Table 8. Metaphor scenarios Metaphor Metaphor units units Scenarios in the scenarios Turn of the Screw Billy Budd Cement Garden
1,021 788 385
83 47 5
149 155 10
Percentage of metaphors belonging to a scenario 14.6% 19.7% 2.6%
In Turn of the Screw and in Billy Budd metaphors occur in more conceptually interwoven sets with a much higher likelihood than in The Cement Garden. For the two 19th-century novellas this indicates complex metaphor use and thus higher “literariness” than would be the case if only conceptually isolated metaphors occurred. 6. Conclusion In summary, an approach to metaphor based on atlas.ti and a complex code system has methodological benefits of various sorts. Qualitative analysis as such emphasizes full coding from the data, instead of looking for preconceived keywords or the like, which adds a degree of text-driven inductive theorizing. Although coding itself is similar to that by hand-written annotations, the software heightens the scope and precision of coding and multiplies retrieval speed by a large factor. Qualitative software is thus highly systematic and can manage a complex research agenda, both of which are preconditions for a comparative analysis of a corpus of several novellas. The multilevel and compositional metaphor coding perspective costs time, but makes up for it with gains in later analysis. First, the type-detail code distinction was created for allowing both a comparative survey analysis (“metaphorizing strategy”) and a content-based detail analysis (“mappings”), respectively. Second, the compositional (source-target) coding approach as such is a precondition for full-text coding of metaphors and allows flexible search queries, including image schema “mesh” that is independent of target domains. Finally, the two-tier coding of metaphoric source domains allows detecting similarities between metaphors at several analytical levels. All in all, the software used here allows handling a multilayered code system, keeping it manageable, and unfolding its analytical power. Screening out distracting codes, merging separately coded texts, or exporting the code apparatus to . Some scenarios of a complex sort were not coded with more than one unit, which explains why the number here is not double or more the mumber of scenarios.
Chapter 13. Metaphors and software-assisted cognitive stylistics
a new text is easy. Furthermore, atlas.ti allows implementing a set of explorative as well as testing tools. It enables hypothesis-building by metaphor distribution graphs (via SPSS for Windows) and generating quantity tables. However, at its core are search queries for codes. Qualitative detail analysis uses them for browsing the quotes of a given mapping. Working out metaphor mappings is equally based on source-target co-occurrences resulting from queries. Finally, quantitative hypothesis testing (as applied for the metaphorizing strategy) also employs queries. I turn now to possible project extensions and future directions. The six-text comparison sketched here, once developed in its full form, will employ atlas.ti to describe relations between subsystems of codes systematically, most importantly concerning how metaphors are embedded in wider narrative structures. This concerns whether, for example dialogues, introspective passages and the telling mode involve different degrees of metaphor use. Other projects may employ the same search tools for exploring interrelations between metaphors and other literary structures such as metaphor and speaker’s mind style, metaphor and narrative perspective, etc. Another extension of the study discussed here will concern theorizing about embodied responses to reading, such as may be created by emotionrelated metaphors, synesthesia and other kinds of imagery. Based on this, it will be inquired which kind of genre is particularly likely to target bodily level reader involvement in the reading process. In closing, let me stress why the presented research agenda, with the application of full-scale and systematic coding, is currently much needed for understanding literary metaphor. The necessity of testing a code system on several texts of some length sharpens the conceptual categories by which we understand literary metaphor. Only a well-defined and broadly applicable code system will allow predicting how text cues and reader response are connected and give rise to specific predictions for the much needed empirical studies. Moreover, the comparative perspective is vital because metaphor functions quite differently in different texts, with the double implication that the method of metaphor analysis cannot be expected to be equally insightful everywhere and that special tool extensions may be needed for particular texts. Full-scale coding is equally important, as it allows the research community to better evaluate the scope of the metaphor methodology in literary theory. Together, these steps will help cognitive linguists to better understand the possibilities and limitations of metaphor analysis and make their case vis-à-vis other fields.
References Danaher, D. 2003. A cognitive approach to metaphor in prose: Truth and falsehood in Leo Tolstoy’s “The Death of Ivan Il’ich”. Poetics Today 24: 439–69.
Michael Kimmel Downes, W. 1993. Reading the language itself: Some methodological problems in D.C. Freeman‘s ‘According to my bond’: King Lear and re-cognition. Language and Literature 2: 121–128. Freeman, D. 1993. ‘According to my bond’: King Lear and re-cognition. Language and Literature 2: 1–18. Freeman, D. 1995. ‘Catch[ing] the nearest way’: Macbeth and cognitive metaphor. Journal of Pragmatics 24: 689–708. Kimmel, M. 2005. From metaphor to the “mental sketchpad”: Literary macrostructure and compound image schemas in Heart of Darkness. Metaphor & Symbol 20: 199–238. Lakoff, G. & Johnson, M. 1999. Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York NY: Basic Books. Lakoff, G. & Turner, M. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago IL: University of Chicago Press. Pragglejazz Group. 2007. MIP: A method for identifying metaphorically used words in discourse. Metaphor & Symbol 22: 1–39. Steen, G. & Gibbs, R.W. 2004. Questions about metaphor in literature. European Journal of English Studies. 8: 337–354. Stockwell, P. 2002. Cognitive Poetics: An Introduction. London: Routledge.
chapter 14
Searching for style in Modern American Poetry David L. Hoover This essay examines a corpus of Modern American Poetry using Zeta and Iota, two recently developed measures of textual difference (Burrows 2007), and some alternatives to them that move beyond the most frequent words of the text, which have been the traditional focus of computational stylistics. These measures concentrate our attention on moderately frequent or rare characteristic words, most of which are content words. These highly characteristic words lead us back to the text, back to questions of interpretation and style, and highlighting such words in texts emphasizes their dense concentration and helps us to visualize how and why intuitive perceptions of stylistic difference are possible. Keywords: computational stylistics, Zeta, Iota, visualization, Modern American Poetry, style
1. Introduction Individual stylistic features are sometimes strikingly apparent, and they have frequently been the focus of perceptive analyses. Often, however, style arises from the cumulative effects of more elusive features distributed throughout the text, and the gap between analysis and interpretation is difficult to bridge persuasively, as has been forcefully and famously (if not always soundly) argued by Fish (1981; see also Simpson, 1993). There have been many responses to Fish (see, for example, Toolan, 1996; Hoover, 2007a), but empirical studies seem among the most promising ways of answering this kind of critique. Probably the best known approaches are the work associated with Willie van Peer’s classic study (1981), the work of Miall and Kuiken (2002; and their website “Reader Response” – http://www.ualberta.ca/~dmiall/reading/index.htm), and the redes project – work discussed in another section of this volume. In this essay I use a
David L. Hoover
different approach that addresses the distributed nature of style by using computational analyses of a large corpus to identify the characteristic vocabulary of some important modern American poets. This approach eliminates from consideration the most frequent words of the texts, some kinds of content words tied closely to setting, character, and plot, and some extremely rare words. Computational stylistics offers powerful, varied, and increasingly popular methods for studying literary style. These methods make it possible to analyze large amounts of text at a level of detail impossible without them. At the same time, however, computational methods tend to focus attention on word lists, statistical patterns, and graphical representations rather than on the text itself. This problem is compounded by the popularity of some very successful methods based on the most frequent words of the text (the, and, of, a, to, in, I, is, that, you, etc.). In his pioneering study of Austen, Burrows (1987) achieves illuminating results by studying very common words (typically the 30–60 most frequent), and his methods continue to produce good results in authorship attribution, style variation, and other investigations involving the classification or discrimination of styles. Yet these words are largely function words lacking any clear relevance to larger thematic and interpretive concerns. Indeed, computational methods often intentionally avoid content words on the assumption that the frequencies of function words are unlikely to be consciously manipulated, and so should be more consistent and stable markers of authorship. Recently, however, there has been a trend toward using more of the word frequency spectrum, and many methods give improved results when the 500–4,000 most frequent words are included (Hoover, 2004, 2007b; van Dalen-Oskam & van Zundert, 2007; Burrows, 2002). Using long word lists has the advantage of including most of the text (the 500 most frequent words typically account for about three-fourths of all the words in a novel). Most of those words are content words obviously relevant to meaning and interpretation, but it is not easy to characterize or discuss thousands of words in ways that illuminate the text. Few stylisticians would now assert an absolute distinction between style and content, but long word lists contain many words too closely related to content to be comfortably included in the characterization of an author’s style (for an excellent discussion of the style-content distinction, see Leech & Short, 1981: 11–41). The high frequency of London, ranch, Tom, Jane, war, or perjury in a text is closely tied to the setting, character, or plot, while the high frequency of must, never, moreover, or whilst is not.
. This kind of work is sometimes called stylometry or corpus stylistics, but “computational stylistics” seems the most appropriate term; “stylometry” tends to suggest mere measurement and corpus stylistics need not be computational in the sense described below.
Chapter 14. Searching for style in Modern American Poetry
2. Zeta and Iota What is needed is a simple method of identifying words that are characteristic of an author or text but are neither among the most frequent words in the language nor too obviously or transparently conditioned by the content. Burrows (2007) discusses two new measures of textual difference, Zeta and Iota, that are specifically designed to focus attention on words that are below the stratum of the most frequent words and that are characteristic of a text or an author (see also Burrows 2005). Although he presents these measures in the context of authorship attribution, their usefulness in identifying an author’s characteristic words is potentially even more useful for stylistic study. Both measures begin with a complete word frequency list for a sample of text by a primary author. The sample is divided into five sections of equal size, and the analyst records how many of these contain each word in the author’s word frequency list. This provides a simple measure of the consistency with which the primary author uses each word. Once this information is collected, it is used to compare the primary author’s sample with samples by other authors and with long poems to be tested for authorship. Sorting the word list on the basis of how many of the primary author’s text sections contain each word allows Burrows to eliminate very frequent words that occur in most texts and concentrate on different parts of the word frequency spectrum. For Zeta, he retains only moderately frequent words – those that occur in three or more of the primary author’s five sections. Where only two poets are being compared, he further reduces the list by removing words that occur more than twice in the works of the second poet. Where many authors are being compared, he removes words that appear in the samples of nearly all of the other authors. Whether there are two authors or many authors, the result is a list of words that are moderately frequent in the primary author and relatively rare in the other author(s). The relative frequency of Zeta words varies with the size of the samples being examined, but in my experiments Zeta words are typically among the 100th to 1,500th most frequent words. . Keyword analysis has a similar goal. Keywords are typically defined as any words for which there is a statistically significant difference in frequency between a text of interest and a reference corpus. The method described above has the advantage that it takes into account each word’s consistency of occurrence in an author’s texts, eliminates the most frequent, is more appropriate for the comparison of two equal samples, and is easy to customize. . Burrows (2007) expands contractions, including a word like can’t in the totals for can and not and tags a few words to distinguish homographs with different grammatical functions. I treat contractions and hyphenated words as single words and do not distinguish homographs. The important thing is to be consistent in any given study.
David L. Hoover
Iota words are rare words, typically occurring only one to four times in the author’s entire corpus, with ranks above 1,500. For Iota, words appearing in more than two of the primary author’s sections are removed. For two authors, the list is limited to words that do not appear at all in the second author’s sample, and, for many authors, the second step removes words that appear in more than about half of the other authors. For a detailed discussion of the calculation of these measures, see Burrows (2007: 30–37); see Hoover (2007b) for an application of them to the style of Henry James. In head-to-head comparisons of Waller and Marvell, Zeta and Iota are remarkably effective in attributing poems as short as 1,000 words to the correct authors in Burrows’s (2007) study, showing that they are capturing important differences between the authors’ styles. When he tests Marvell and Waller against the 24 other primary samples, twenty-four independent poems by the primary authors, and twenty-one poems by other authors, Iota works very well based on either author’s word list, and Zeta works well based on Waller’s word list. Zeta based on Marvell’s word list produces failures that Burrows (2007) suggests may result from the contrast between the political satires being tested and the largely pastoral nature of Marvell’s other poetry. Zeta and Iota will require further testing and possibly some modification before they can be confidently applied to authorship problems, but Burrows reports that he has never seen them fail in a head-to-head comparison (2007: 43). Here I will concentrate on what seems a more important characteristic: Zeta and Iota allow us to concentrate on a relatively small subset of words that are stylistically characteristic of an author. These characteristic words (primarily content words) lead back to the text, and allow us to begin a new kind of investigation into the style of content and the content of style. This new investigation tests Burrows’s methods on modern American poetry, shifting both continent and century. I began by downloading samples (ranging from about 24,000 to 138,000 words) of poetry by 25 important modern American poets born between 1869 and 1913 from Literature Online (gateway.proquest. com) through NYU’s Bobst library. I regularized dashes and removed bibliographic information, notes, and other extraneous information. Then (paralleling Burrows’s investigation) I removed the following long poems (the shortest about 900 words) from the primary samples so that I could test whether Zeta and Iota could correctly attribute them to their authors: Frost’s “The Housekeeper,” “Snow,” and “A Servant to Servants,” and Stevens’s “The Man with the Blue Guitar,” “Notes Toward a Supreme Fiction,” “The Auroras of Autumn,” and “Sunday Morning.”
. I first selected all important poets with large enough samples of poetry available electronically, then limited the range of birth dates as much as possible while including 25 poets.
Chapter 14. Searching for style in Modern American Poetry
Head-to-head tests of Stevens vs. Frost and Frost vs. Stevens give even more definitive results than Burrows (2007) achieves for Marvell vs. Waller. Zeta and Iota easily distinguish the two poets, whichever poet’s word list is used, and increasing the stringency of the stipulations produces fascinating results. The 63 words appearing in at least three of the five sections into which Frost’s sample was divided but not at all in Stevens’s five sections and the 125 words appearing in all five of Stevens’s sections but not at all in Frost’s five sections are excellent potential markers of the styles of the two poets. The lowest frequency of Frost’s characteristic words in any of his sections or independent poems is almost fifty times as great as in any of Stevens’s sections or independent poems, and the lowest frequency of Stevens’s characteristic words in any of his sections or poems is almost six times as great as in any of Frost’s sections or independent poems. Both groups of words range in rank within their word lists from about the 100th to the 1,400th most frequent, and occur with frequencies of roughly once in 1,000 words to once in 10,000 words. Furthermore, Frost’s 63 characteristic words occur more than 700 times in his sections and poems but never in Stevens’s sections and only twice in his independent poems, though my corpus includes only about 22,000 words of Frost’s poetry compared to about 70,000 of Stevens’s. Stevens’s 125 characteristic words occur more than 2,300 times in his sections and poems, but never in Frost’s sections and only 19 times in his independent poems. These two sets of marker words return our attention to the texts. 63 Characteristic Frost words rare in Stevens: anyway, aren’t, asked, axe, bark, barn, birch, bow, boy’s, can’t, cellar, climbing, couldn’t, dared, didn’t, dragged, driven, farmer, folk, folks, fool, foolish, front, hadn’t, haven’t, he’d, he’ll, he’s, hope, hurt, I’d, I’ll, I’ve, isn’t, knoll, letter, likely, meadow, mistake, mustn’t, nearly, o’clock, pasture, paused, pay, scared, sell, shouldn’t, somewhere, store, straining, strangeness, stranger, surely, swinging, they’re, try, wasn’t, ways, wouldn’t, you’d, you’ll, you’ve 125 Characteristic Stevens words rare in Frost: actual, also, animal, archaic, attend, autumn, autumnal, bells, blank, blood, book, brings, bronze, café, calm, casual, central, centre, chaos, clearest, closely, colors, composed, constant, cries, crown, cry, deeply, dressed, ear, edge, element, escape, essential, experience, false, farewell, fiery, filled, final, fortune, fragrance, freedom, genius, ghosts, glistening, glittering, gold, grew, hears, honey, horizon, hymns, ideas, image, images, imagination, imagined, immaculate, immense, indifferent, inhuman, invisible, lesser, lights, lover, luminous, march, minor, misery, moonlight, motions, mountains, nakedness, nameless, nothingness, ocean, page, perfection, poem, poems, poet, poetry, point, possible, poverty, profound, queen, quick, reading, reads, reality, relation, returning, rise, rising, rock, season, secret, self, selves, senses, shapes, shore, singular, skies, solitude,
David L. Hoover
sovereign, spirit, stands, state, strong, syllable, syllables, thoughts, thunder, total, universal, visible, vivid, voluble, walks, waves, winter’s, written Frost’s list is striking for the prevalence of contractions and the high concentration of Anglo-Saxon vocabulary with a rural flavor, characteristics in accord with his reputation for everyday language and his heavy use of dialogue. Stevens’s list, on the other hand, is more abstract, more formal, and more Latinate in origin, as befits a “difficult” poet, and his penchant for writing poems about poetry is reflected in the many words related to speech, writing, poetry, and the imagination. Considering how short Frost’s list is, it contains a remarkable number of families of related words: folk/folks, fool/foolish, and strangeness/stranger, and Stevens’s longer list contains autumn/autumnal, central/centre, cries/cry, image/images/imagination/ imagined, poem/poems/poet/poetry, reading/reads, rise/rising, self/selves, and syllable/ syllables. Such word families provide further evidence that we are dealing with truly characteristic vocabulary (see Hoover 2007b: 184–85 on word familiar in James). By highlighting these two lists of words in the texts of the authors, we can see just how characteristic they are. Obviously, their density varies a great deal, and some passages lack them entirely, but the brief passages below show just how densely even these few words sometimes cluster (the percentage of all words that are Zeta words is indicated at the end of each passage). Frost, from “The Death of the Hired Man” (63 Zeta words) “You needn’t smile – I didn’t recognise him – I wasn’t looking for him – and he’s changed. Wait till you see.” “Where did you say he’d been?” “He didn’t say. I dragged him to the house, And gave him tea and tried to make him smoke. ....................................... “I’d not be in a hurry to say that.” “I haven’t been. Go, look, see for yourself. But, Warren, please remember how it is: He’s come to help you ditch the meadow. He has a plan. You mustn’t laugh at him. 13% Stevens, from “Dutch Graves in Bucks County” (125 Zeta words) These violent marchers of the present, Rumbling along the autumnal horizon, Under the arches, over the arches, in arcs Of a chaos composed in more than order, March toward a generation’s centre. 19%
Chapter 14. Searching for style in Modern American Poetry
Stevens, from “A Primitive like an Orb” (125 Zeta words) The central poem is the poem of the whole, The poem of the composition of the whole, The composition of blue sea and of green, Of blue light and of green, as lesser poems, And the miraculous multiplex of lesser poems, Not merely into a whole, but a poem of The whole, the essential compact of the parts, The roundness that pulls tight the final ring 17%
It is important to emphasize the frequency of Zeta contractions in the Frost passage. Obviously, such common contraction as he’s, didn’t, wasn’t, I’d, and haven’t only characterize Frost compared with Stevens and other poets who rarely use them. Poets who frequently use dialogue or cultivate a casual style use more of these words, and their poetry is distinguished from Frost’s by quite different words. This reminds us that studying style is always a comparative undertaking: no feature can be striking or characteristic unless it differs from some norm or imagined alternative. Stipulations more like those Burrows (2007) uses in distinguishing Marvell and Waller identify a much larger set of distinctive Frost and Stevens words. Frost vs. Stevens 190 Zeta words Consistency stipulation: Frequency stipulations:
found in a minimum of 2 of Frost’s sections a minimum total frequency of 3 in Frost’s sections a maximum total frequency of 1 in Stevens’s sections
221 Iota words Consistency stipulation: Frequency stipulations:
found in a maximum of 2 of Frost’s sections a total frequency of exactly 2 in Frost’s sections not present in Stevens’s sections
Stevens vs. Frost 834 Zeta words Consistency stipulation: Frequency stipulations:
found in a minimum of 3 of Stevens’s sections not present in Frost’s sections
. Although Burrows (2007) includes hapax legomena (words occurring only once) in his Iota words, I focus here on words appearing at least twice, assuming that they are more characteristic of the authors. Stipulating a frequency of exactly two in Frost eliminates the hapax legomena and prevents any overlap between the Iota and Zeta lists (the latter with a minimum frequency of three).
David L. Hoover
1968 Iota words Consistency stipulation: Frequency stipulations:
found in a maximum of 2 of Stevens’s sections a minimum total frequency of 2 in Stevens’s sections not present in Frost’s sections
Consider the frequencies of these characteristic words in two long poems. Zeta and Iota words together account for 160 (11%) of the 1509 words (tokens) and 94 (17%) of the 560 different words (types) in Frost’s “A Hundred Collars.” For Stevens’s “Sunday Morning,” the concentration is even higher: 144 (17%) of the 867 tokens and 121 (28%) of the 430 types are either Zeta or Iota words. Here are two passages from the poems. Frost, from “A Hundred Collars” (Frost Zeta words, Frost Iota words) Lancaster bore him – such a little town, Such a great man. It doesn’t see him often Of late years, though he keeps the old home-stead And sends the children down there with their mother To run wild in the summer – a little wild. Sometimes he joins them for a day or two And sees old friends he somehow can’t get near. They meet him in the general store at night, Pre-occupied with formidable mail, Rifling a printed letter as he talks. They seem afraid. He wouldn’t have it so: Though a great scholar, he’s a democrat, If not at heart, at least on principle. Lately when coming up to Lancaster His train being late he missed another train And had four hours to wait at Woodsville Junction After eleven o’clock at night. . . . 14% Zeta and Iota words Stevens, from “Sunday Morning” (Stevens Zeta words, Stevens Iota words) She hears, upon that water without sound, A voice that cries, “The tomb in Palestine Is not the porch of spirits lingering. It is the grave of Jesus, where he lay.” We live in an old chaos of the sun, Or old dependency of day and night, Or island solitude, unsponsored, free, Of that wide water, inescapable. Deer walk upon our mountains, and the quail Whistle about us their spontaneous cries;
Chapter 14. Searching for style in Modern American Poetry
Sweet berries ripen in the wilderness; And, in the isolation of the sky, At evening, casual flocks of pigeons make Ambiguous undulations as they sink, Downward to darkness, on extended wings. 25% Zeta and Iota words
These passages reveal one unfortunate characteristic of Iota: words that are infrequent in one poet and absent from the other often include proper names like Lancaster (twice) in the Frost passage and Palestine, and Jesus in the Stevens passage. The only remedy is to remove obvious proper nouns manually, realizing that this is an error-prone and somewhat subjective task. Frost’s Zeta words above are remarkable in containing contractions and rather homey nouns and verbs like store, letter, train, keeps, talks, and missed. Stevens’s Zeta words include specific concrete nouns and verbs (tomb deer, quail, mountains, pigeons, hears, cries, whistle), interspersed with contrasting abstract and largely Latinate nouns and adjectives (spirits, chaos, solitude, isolation, undulations, inescapable, spontaneous, casual, extended). 3. A simple alternative to Zeta and Iota for one-to-one comparisons For comparisons between the styles of two poets, simpler measures than Zeta and Iota are quite effective. For example, making a word list of each poet’s sample as a whole and then selecting words at least three times as frequent in one poet as in the other gives a very good contrastive picture, even without the additional information about the consistency of use that is present in Zeta and (to some extent) in Iota. Switching from Frost to Edwin Arlington Robinson (1869–1935), a poet with a sample even larger than Stevens’s, and manually deleting proper names from the lists yields 1,322 words used at least three times as frequently by Robinson as by Stevens and 3,983 words used at least three times as frequently by Stevens as by Robinson. Here is the passage from Stevens’s “Sunday Morning” (quoted above) with both sets of words marked. (Stevens 3X Robinson, Robinson 3X Stevens) She hears, upon that water without sound, A voice that cries, “The tomb in Palestine Is not the porch of spirits lingering. It is the grave of Jesus, where he lay.” We live in an old chaos of the sun, Or old dependency of day and night, Or island solitude, unsponsored, free, Of that wide water, inescapable.
David L. Hoover
Deer walk upon our mountains, and the quail Whistle about us their spontaneous cries; Sweet berries ripen in the wilderness; And, in the isolation of the sky, At evening, casual flocks of pigeons make Ambiguous undulations as they sink, Downward to darkness, on extended wings. 30% Stevens 3X Robinson words
In spite of my deletion of Palestine and Jesus from the list, the density of characteristic Stevens words is even higher than it was above. This method also marks most of the same words, suggesting that these words are characteristic of Stevens in general, not just in comparison with Frost or Robinson. In the following complete short poem, the characteristic Stevens vocabulary is even denser. Stevens, “The Candle a Saint” (Stevens 3X Robinson, Robinson 3X Stevens) Green is the night, green kindled and apparelled. It is she that walks among astronomers. She strides above the rabbit and the cat, Like a noble figure, out of the sky, Moving among the sleepers, the men, Those that lie chanting green is the night. Green is the night and out of madness woven, The self-same madness of the astronomers And of him that sees, beyond the astronomers, The topaz rabbit and the emerald cat, That sees above them, that sees rise up above them, The noble figure, the essential shadow, Moving and being, the image at its source, The abstract, the archaic queen. Green is the night. 38% Stevens 3X Robinson words
In these poems by Stevens, note that almost all of the highlighted words are content words. Concrete nouns like deer, mountains, pigeons, berries, rabbit, cat, evening, night, and the adjective green are clearly related to the physical setting of the poems, but their mixture with abstract nouns like chaos, solitude, and isolation, and adjectives like spontaneous, unsponsored, apparelled, essential, abstract, noble, and archaic suggest that Stevens is no simple nature poet. His “casual flocks of pigeons make/Ambiguous undulations,” his night is “green kindled and apparelled,” his rabbit is topaz, and his cat is emerald. The world of the imagination seems as important as the natural world.
Chapter 14. Searching for style in Modern American Poetry
Here are two complete short poems by Robinson for comparison (the markings for the Stevens and Robinson words have been reversed): Robinson, “Twilight Song” (Robinson 3X Stevens, Stevens 3X Robinson) Through the shine, through the rain We have shared the day’s load; To the old march again We have tramped the long road; We have laughed, we have cried, And we’ve tossed the King’s crown; We have fought, we have died, And we’ve trod the day down. So it’s lift the old song Ere the night flies again, Where the road leads along Through the shine, through the rain. Long ago, far away, Came a sign from the skies; And we feared then to pray For the new sun to rise: With the King there at hand, Not a child stepped or stirred – Where the light filled the land And the light brought the word; For we knew then the gleam Though we feared then the day, And the dawn smote the dream Long ago, far away. But the road leads us all, For the King now is dead; And we know, stand or fall, We have shared the day’s bread. We may laugh down the dream, For the dream breaks and flies; And we trust now the gleam, For the gleam never dies; – So it’s off now the load, For we know the night’s call, And we know now the road And the road leads us all. Through the shine, through the rain, We have wrought the day’s quest; To the old march again
David L. Hoover
We have earned the day’s rest; We have laughed, we have cried, And we’ve heard the King’s groans; We have fought, we have died, And we’ve burned the King’s bones, And we lift the old song Ere the night flies again, Where the road leads along Through the shine, through the rain. 31% Robinson 3X Stevens words Robinson, “An Old Story” (Robinson 3X Stevens, Stevens 3X Robinson) Strange that I did not know him then, That friend of mine! I did not even show him then One friendly sign; But cursed him for the ways he had To make me see My envy of the praise he had For praising me. I would have rid the earth of him Once, in my pride … I never knew the worth of him Until he died. 36% Robinson 3X Stevens words
Though there are far fewer characteristic words for Robinson than Stevens, their density in the highlighted passages is quite similar. Yet their character is radically different. Function words like have, him, we’ve, me, it’s and the adverbs now, and again are quite frequent in English generally, but Robinson uses them at least three times as frequently as Stevens. In “Twilight Song,” note how the present perfect tense of the narrative (contrasting with Stevens’s more lyical present tense) conditions the frequency of have. The content words of Robinson’s poems also tend to be much less specific and less striking, with nouns like road, load, king, groans, gleam, but also “moral” words like envy, praise, pride, and worth, and verbs like laugh(ed), lead(s), share, know/knew, but also feared, fought, died, and cursed. (For a more general discussion of nouns in Stevens and other poets, see Hoover, 2006: 80–87). It would be fair to characterize Robinson’s style as more grammatical and Stevens’s as more lexical. These brief comments on the characteristic vocabularies of Frost, Stevens, and Robinson can only suggest some of the large differences among poetic styles, but a fuller analysis of longer marked passages by these or any other poets would surely
Chapter 14. Searching for style in Modern American Poetry
be an effective way of investigating their stylistic characteristics and differences, especially when coupled with part-of-speech tagging. Note also that the lists of characteristic words can be tuned by selecting words with even larger differences, reducing them to a more manageable size (about 250 words are ten times as frequent in Robinson as in Stevens; about 250 are twenty times as frequent in Stevens as in Robinson). 4. A simple alternative to zeta and iota in one-to-many comparisons In addition to identifying vocabulary that can characterize two poets with respect to each other, computational methods can also identify vocabulary that distinguishes a single poet from a large group of other poets. As described above, Zeta and Iota often correctly attribute individual long poems in the presence of many other poets and independent poems, but intensive testing suggests that they not always effective, and Iota’s inclusion of hapax legomena seems likely to be more effective for classification than for discovering characteristic vocabulary. The method I present here begins with word frequency lists for the samples of poetry by all 25 of the American poets mentioned above. Isolating the word list for Stevens allows the frequencies of all the words in his sample to be compared with their frequencies for the other 24 poets. After calculating the maximum frequency for each word in the entire set of poets and comparing it with the frequency of the word in Stevens, I select words at least 1.5 times as frequent in Stevens as in any other poet. (The cut-off is to some extent arbitrary, and is chosen to produce a manageable set of words.) I add all words that occur at least twice in Stevens but not in any of the other poets and manually remove the proper names, as above. Repeating this process for three other poets results in 655 characteristic words for Stevens, 245 for Robinson, 324 for T. S. Eliot, and 489 for Carl Sandburg. Table 1 has three columns of these words for each of the four poets. I have temporarily concealed the identities of the poets, so that readers can try to identify them before reading the discussion below the table. The first column for each poet is arranged in descending order by the word’s rank in the entire set; for example, time, in the first column for Poet 1, ranks 71st in the entire set, and for, in the first column for Poet 2, ranks 13th in the entire set; the rest rank progressively lower. The second column for each poet lists words with frequencies that exceed the maximum for all other poets by the greatest margin; for example, the frequency of window-panes in Poet 1 is eleven times its maximum frequency in any of the other 24 poets, and the frequency of expectancy in Poet 2 is nearly five times its maximum in any of the others. The third column for each poet lists words appearing at least twice in his sample but not in any of the other poets.
David L. Hoover
Table 1. Characteristic words of four poets
Poet 1
Poet 2
Most Freq. Highest Ratio
Most Freq. Most Freq. Highest Restr. Ratio
Most Freq. Restr.
time between end word meet forgotten future beginning afternoon church desert build waste prayer stair action kingdom experience purpose stillness pattern movement tea swell service
goonight yew-tree twit transcript rose-garden guesses annunciation woodthrush wainscot waggons unstilled tumid transitory timekept teacups situations revisions ‘potamus pocketed pluckest lifetime’s juniper-tree hippo’s hedgerow formulated
unhappily pernicious hautboys ascendency regretfully admonition wizardry vagaries urbanity unprofitable soever schisms recreance played-out misbegotten jeopardized irretrievably iniquities impending gratefulness forlornly divination delving credulity completeness
an without mind part clouds themselves sense except poem sounds weather forms
window-panes reappears horoscope unspoken restoring committees street-lamp relight misunderstanding undefeated crumpets vegetation imprecision divan prickly aunts yew voyagers satisfactory revisit quickens nymphs horses’ rats’ loitering
for have there had more than may made might found king knows told friend longer doubt smiled hers forgive ruin fancy feared regret wondered fearing
Poet 3 external fictive cadaverous guitar realist personage panache venerable seemings profounder rhapsody voluminous
expectancy specks wondered frowned faggots accordingly whatsoever knight assuredly confirmation wormwood prison-yard hungrily fortified determining wayside foiled smiled doubts scanned told entertainment albeit reptile heretofore Poet 4
semblables transparence bethou secondary meta-men rejects slopping c’était exactest residuum secrete acutest
on men red along women dust ten cool boys laughter steel mist
mockingbird buckles roan gamblers post-office hoboes harbor’s passers-by wagons khaki slang hustlin’
cheep cornhuskers smokestacks gigglers flimmering blithery tiddy teamster summer-white sluggers newsies knucks (Continued)
Chapter 14. Searching for style in Modern American Poetry
Table 1. Continued
Poet 3
Poet 4
Most Freq. Highest Ratio
Most Freq. Most Freq. Highest Restr. Ratio
Most Freq. Restr.
becomes rudiments feeling spouse bloom seraphs hero regalia idea radial one’s promenades figure evolved central cockade imagination orator object bodiless ideas meditation tragic artifice blank origin
armorer’s barer canon encore evocations foyer ha-e icon illustration nougats pantaloons plural undulation
iron-jawed hunky handler falltime earful dustpan dago concertina chuff-chuff bricklayer bobsled anod ac-ci-dents
heads horses lines million dusk shoulders ready corn valley brothers six hunger guns
bastions horseback good-by fixes northwest mob oval stub scabs policeman playthings gloaming jag
Surely most readers familiar with his poetry will recognize Sandburg as Poet 4. The distinctive and almost indexical word cornhuskers, in combination with words like steel, shoulders, harbor’s, and smokestacks, makes this a particularly characteristic list. Although bethou, semblables, and meta-men in Poet 3 may be indexical for some readers, identifying this as Stevens, his list seems deeply characteristic throughout, especially some of the high frequency words, such as mind, clouds, sense, poem, sounds, idea(s), central, and imagination, in the first column. The other two lists seem less distinctive. Many of the words in the lists for Poet 1, are quite rare, appearing only a few times in his entire sample (the smallest of the four samples discussed here, about 21,000 words), though anyone who has read Eliot’s “The Hippopotamus” recently will probably find the presence of both hippo’s and ‘potamus indexical, and the same might be said for transcript from his “The Boston Evening Transcript.” For E. A. Robinson, Poet 2, the difficulty is that much of his distinctive vocabulary is also relatively frequent overall. In spite of having frequencies at least 1.5 times the maximum for any other poet, the first 15 of the words in his first column are also among the 500 most frequent for the entire set. Robinson’s words with the highest ratio of occurrence in his sample versus the maximum for any other poet in his second column are also unusual: other poets’ lowest ratios are nearly as high as Robinson’s highest ratios. Thus, the distinctiveness of Robinson’s style is more quantitative than qualitative, more a matter of using common vocabulary at uncommonly high frequencies, and is perhaps more subtle.
David L. Hoover
5. Conclusion Further investigation is needed into how and to what extent readers are able to recognize a previously unread text by an author they know well. The interplay between differences based on relative frequency and those based on the presence of large numbers of unique or indexical words seems especially interesting. The situation is so complex that it may be difficult or even impossible to design valid experiments with human subjects. Perhaps, however, methods of extracting verifiably “characteristic” vocabulary computationally can in itself provide some insight, along with suggesting words for more directly empirical investigation. Marking the positions of such words in an author’s texts can help us to visualize the density of stylistic markers and thereby understand the ability of readers to recognize previously unread samples from an author they know well. Computational methods that focus on very frequent words (or even, as has recently become popular, frequent sequences of letters) may be effective in correctly attributing texts to their authors, but we also need methods that focus on stylistically characteristic words. Such words, with their connections to both style and content, help to focus our attention on the central literary and aesthetic questions of style, interpretation, and meaning.
References Burrows, J.F. 1987. Computation into Criticism. Oxford: Clarendon Press. Burrows, J.F. 2002. ‘Delta’: A measure of stylistic difference and a guide to likely authorship. Literary and Linguistic Computing 17: 267–287. Burrows, J.F. 2005. Who wrote Shamela? Verifying the authorship of a parodic text. Literary and Linguistic Computing 20: 437–450. Burrows, J.F. 2007. All the way through: Testing for authorship in different frequency strata. Literary and Linguistic Computing 22: 27–47. Fish, S.E. 1981. What is stylistics and why are they saying such terrible things about it? In Essays in Modern Stylistics, D.C. Freeman (ed.), 53–78. London: Methuen. Hoover, D.L. 2004. Delta prime? Literary and Linguistic Computing 19: 477–495. Hoover, D.L. 2006. Hot-air textuality: Literature after Jerome McGann. Text Technology 15: 75–107. Hoover, D.L. 2007a. The end of the irrelevant text: Electronic texts, linguistics, and literary theory. Digital Humanities 1(2); online: http://www.digitalhumanities.org/dhq/vol/001/2/000012. html. Hoover, D.L. 2007b. Corpus stylistics, stylometry, and the styles of Henry James. Style 41: 174–203. Leech, G.N. & Short, M.H. 1981. Style in Fiction. New York NY: Longman. Miall, D.S. & Kuiken, D. 2002. The effects of local phonetic contrasts in readers’ responses to a short story. Empirical Studies of the Arts 20: 157–175.
Chapter 14. Searching for style in Modern American Poetry
Simpson, P. 1993. Language, Ideology and Point of View. London: Routledge. Toolan, M. 1996. Stylistics and its discontents; or, getting off the Fish ‘hook’. In The Stylistics Reader: From Roman Jakobson to the Present, J.J. Weber (ed.), 117–135. London: Arnold. Van Dalen-Oskam, K. & van Zundert, J. 2007. Delta for Middle Dutch – Author and copyist distinction in Walewein. Literary and Linguistic Computing 22: 345–362. Van Peer, W. 1981. Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm.
chapter 15
The laws governing the history of poetry Colin Martindale Poetry and the other arts by definition require the production of novel artifacts. Someone who merely copies a poem or plays a piece of music is not thought of as a poet or composer. Even if this were not the case, we habituate to the repetition of the same thing and want something new. The pressure for novelty may be a major consideration or a nuisance for poets, but it has exerted a constant pressure since poetry was first written whereas other pressures have come and gone. When we consider how novel ideas are produced, we see that the pressure for novelty dictates not only that poetry will change across time but that it will change in a very specific direction. A theory describing how entropy must increase in poetry and what sorts of contents and styles must be found in the history of any poetic tradition is described. A content analysis of samples of British poetry by 170 poets born between 1290 and 1949 is described. Support for all of the theoretical predictions was found. Keywords: history of poetry, British poetry, sociocultural evolution, novelty, empirical study
1. Introduction Across the last several decades, we have seen great strides in the empirical study of literature and the other arts. The idea of studying literature in a scientific manner had been mentioned but very seldom implemented before the 1960s. One important trend has been the development of experimental reader-response approaches that allow us to see what ordinary readers think about what they read rather than relying upon the conjectures of literary critics as to what they might possibly be thinking. Of course, Willie van Peer has been in the forefront of this line of inquiry. On another hand, we have seen the development of computerized content analytic methods (e.g., Stone 1966). These have allowed us to reduce huge amounts of text
Colin Martindale
to a manageable set of quantified measures so that we can see the forest and not be distracted by the trees. In this chapter, I describe a general theory of the development of cultural artifacts as it applies to the history of poetry. In testing the theory as instantiated in the case of poetry, it was necessary to rely heavily upon computerized content analytic techniques. It should be noted that the theory applies to all of the arts. Other methods that need not concern us here are needed in order to test the theory as it applies to, say, painting or music. All of the arts necessarily develop in the same manner; only the details differ. I have presented the theory of poetic history in a quasi-Darwinian framework. This is merely a temporary expedient.
2. Need for novelty Whatever else they must be, works of art and literature must be novel. If I copy a poem already written by someone else, no one will think of me as being a poet. Many theorists have pointed out that if art and literature are characterized by traits such as novelty and disruption of expectation, a necessity for change is built into them. If a work of art must be novel, each successive work of art must be different from all prior works, or it will not qualify as a work of art at all. The Russian Formalists and Czech Structuralists argued that poetic devices involve “estrangement” or “deformation.” What gives poetry its effect is the use of words in ways that are unusual or unexpected. The deformed word usages in poetry intensify perception and attract attention. With repetition, the Formalists argued that linguistic deformations and estrangements gradually become “automatized” (Tynjanov 1929). They lose their effect. Several Formalist theorists such as Shklovsky (1919) and Mukařovský (1940) derived from this fact the hypothesis that literature must necessarily evolve. If aesthetic effects arise from deformations, and if deformations are gradually automatized, then there is a constant pressure on successive artists to produce new deformations.
3. Effects of repetition on aesthetic preference According to Berlyne (1971), preference for any stimulus is based upon the impact value of that stimulus. The impact value of a stimulus is determined by collative properties (e.g., novelty, complexity, surprisingness, unpredictability), ecological properties (signal value or meaning), and psychophysical characteristics in some sense intrinsic to the stimulus (e.g., size, intensity, pitch). Berlyne’s theory
Chapter 15. Poetic History
of aesthetic preference was wrong (Martindale, Moore & Borkum 1990), but it is useful to retain his notion of the three aspects of stimuli that evoke pleasure. Repeated presentation of a given work can either increase or decrease preference for it. This is because presentation of any stimulus both sensitizes and fatigues the mental representation of it; fatigue dissipates more rapidly than sensitization (Martindale 2007a). Thus, the Formalist theorists were not correct in their contention that literary devices would in all cases become automatized and cause a pressure for change. Distributed repetition leads to what is called the mere exposure effect (Zajonc 1968). The more frequently a stimulus has been encountered, the better people like it so long as it is presented interspersed with a number of other stimuli. For example, if one goes to the theater or concerts rather regularly, he or she will most prefer the plays or symphonies that he has encountered most frequently before. Thus, we are likely to prefer Beethoven’s 5th Symphony to his 8th Symphony, at least in part because we have heard the former more often than the latter. On the other hand, with massed repetition (repetition of the same or very similar stimuli over and over again), we find habituation. Fatigue overwhelms sensitization, and we like the stimulus less and less with each repetition (Martindale 2007a). There is a mass of experimental evidence for the occurrence of habituation with stimuli in general and aesthetic stimuli in particular. Let the reader imagine that he or she is locked in a room and forced to listen to Beethoven’s 5th Symphony over and over for 10 hours or so. It is easy enough to understand that what one once loved, he or she will come to hate: one is going to develop an increasing distaste for this cursed piece of noise. If one doubts this, this need not remain a thought experiment but can easily enough be done on oneself. The Formalist theorists were thinking of the more intuitively obvious habituation rather than of the mere exposure effect when they argued that a pressure for change is intrinsic to art and literature. Which effect is found depends upon frequency of exposure. The creators of art and literature are more or less constantly exposed to works in the genre in which they are working. Thus, we should expect them to show habituation and a need for novelty. The extent to which an external audience exhibits habituation vs. the mere exposure effect will depend upon the frequency with which it is exposed to works in a given genre. An interesting thing about the high arts is that there is not much of an external audience. Poetry is of special interest, because poets have more or less always written for each other and, to a lesser extent, for literary critics. If anyone beyond this closed circle happens to read their poetry, it is of little or no concern to them whether the person likes what they have written. The poetry has not been written for such casual readers.
Colin Martindale
4. Compensating for habituation If a series of poets kept producing the same or very similar works of art, liking for their productions would decrease across time. To compensate for this habituation, it is necessary for successive poems to have more and more impact value. In principle, this could be accomplished by manipulating any of the components of what makes art interesting. In a medium such as literature, it is impossible to compensate for habituation by increasing psychophysical properties. One could print larger books or books printed in odd fonts, but if this worked at all, it could hardly work for very long. Attention can also be increased by increasing the meaningfulness of an artistic work. However, people vary widely in what is significant to them. Poets cannot be sure that what is more important for them will also be of interest for their audience if they even have one. Historically, there has been a danger in creating art that is too meaningful. Modern western politicians do not care what poets write about, as virtually no one except other poets reads what they write. However, more authoritarian régimes do take an interest in poetry and the other arts. If they do not like what a poet has written, the poet can end up exiled or dead. In any event, poetry is not really suited to convey material of any complexity. Erasmus Darwin (1789) did write a botanical treatise in verse. It has always been regarded as an odd curiosity in which the verse interfered with clear expression of his ideas, and the ideas spoiled the verse. Laymen and amateur poets have the notion that poetry is the proper medium in which one can express this or that deep emotion. This was true centuries ago, but references to emotions in at least British, French, and American poetry have systematically declined for the last several centuries. There are significantly more references to emotions on the front page stories in the New York Times than in contemporary poetry (Martindale 1990). The reason is not far to find. Everything worth saying about emotions has already been said. Keep in mind that repetition is not allowed in poetry. If a modern poet says anything about love, he would be repeating something already said hundreds of years ago. On the other hand, collative properties such as novelty or unpredictability or surprisingness are much freer to vary in poetry and all of the arts. There are limits on how large a painting can be or how loud music may be played. The meaning of music is rather indeterminate, and it is not very clear how an artist could increase the meaningfulness of a painting of a tree or of a nude woman. Thus, the necessity to increase the impact value of aesthetic products over time eventually comes down to a pressure to increase novelty, incongruity, unpredictability, and other collative variables. Another way of putting this is that the Second Law of Thermodynamics applies to the art world just as to the physical world (Martindale 1995b): Entropy, disorder, or unpredictability must always increase and can never decrease.
Chapter 15. Poetic History
5. Hedonic selection in aesthetic evolution The criterion in aesthetic evolution is analogous to Darwin’s (1871) idea of sexual or hedonic selection rather than to his more well known criterion of “fitness” to the environment first proposed in 1859. Both criteria operate on artistic products, but their effects are quite different. Selection on the basis of preference has been present ever since works of art were first produced. Habituation is a universal property of nervous tissue. Thus, hedonic preference has exerted a constant pressure in the same direction throughout the entire course of human history. On the other hand, social “fitness” has varied wildly across time. Taine (1863) proposed a theory of art history based explicitly upon the adaptationist theories of Darwin (1859). It is of little more than historical interest today, as it does not account well for the data of art history. The main problem is that the social environment is prone to rather quick and dramatic changes. In 1936, it may have struck a German architect that swastikas would make nice decorations for his buildings. However, the Thousand Year Reich did not last quite as long as promised. A mere 10 years later at best the decorations had all been removed, and at worst the entire building had been destroyed if it did not look quite right to the occupying powers. Similarly, pornography has low fitness in a puritanical society, moralistic literature has low fitness in a licentious society, and so on. It is difficult for an artist to know when the tide is going to turn. Thus, fitness has not exerted a consistent, unidirectional pressure on works of art. I am not asserting that artists are motivated solely by a quest for novelty. They are interested in accomplishing many other things besides making their works novel. However, what these other things are varies unsystematically, whereas the pressure for novelty is unremitting, constant, and consistent. Thus, only it can produce systematic trends in artistic form and content. This is true even if need for novelty is a comparatively unimportant motive or a downright nuisance for any specific artist. Across long periods of time, it would appear that all forces, save the pressure for novelty, acting on artists randomize out. Formalist theorists such as Mukařovský (1940) and Tynjanov (1929) agreed that their evolutionary theory could not explain the direction of aesthetic changes, that it is necessary to look to extra-artistic social or cultural forces for any such explanation.
6. Means of production of novel ideas In fact the direction in which any art must change is intrinsic to the art producing system. There is no need to look for extra-artistic causes (Martindale 1975, 1990, 2007c).
Colin Martindale
The pressure for novelty not only produces continuously more novelty but leads to predictions as to the direction that art history must move. These predictions arise from a consideration of the psychological means whereby works of continually increasing novelty could possibly be produced. How do successive poets produce poetry that becomes more and more novel, original, or incongruous over time? To answer this question, we must ask how novel ideas are produced in the first place. According to Kris (1952), original ideas arise from a biphasic process. An initial inspirational stage involving “regression” to a primitive, associative form of thought is followed by a subsequent stage of elaboration with a relatively less regressed mode of thought. All plausible theories of creativity are variants of Kris’s theory or it is a variant of them (Martindale 1981, 2007b). By regression is meant a movement from conceptual thinking toward primordial thought. The conceptual-primordial continuum is the fundamental axis along which states of consciousness and types of thought vary (see Fromm 1978). Freud’s (1900) primary process versus secondary process continuum is but one of many descriptions. Some of the other well-known divisions are Hobbes’s (1642) unguided versus designed thought; Vico’s (1730/1744) poetic versus rational thinking, Nietzsche’s (1872) Dionysian versus Apollonian states; Wundt’s (1896) associative versus intellectual thinking; Lévy-Brühl’s (1910) prelogical versus scientific thinking; Cassirer’s (1925) mythic versus rational modes; Jung’s (1935) eros versus logos; Piaget’s (1936) autistic versus logical: Goldstein’s (1939) concrete versus abstract attitudes; Werner’s (1948) dedifferentiated or syncretic versus differentiated thought; McKellar’s (1957) A-thinking versus R-thinking; Maslow’s (1957) B-cognitian versus A-cognition; Berlyne’s (1965) autistic versus directed thinking; Neisser’s (1967) pre-attentive processes versus focal attention; Bogen’s (1969) appositional versus prepositional thought; Klinger’s (1971) respondent versus operant sequencing; Bruner’s (1991) narrative versus paradigmatic modes; and Petrov’s (2003) right-hemisphere versus left-hemisphere thinking. They focused upon rather different aspects of the continuum but in the main agreed as to what changes as we move from one end of the continuum to the other. Conceptual cognition is abstract, logical, purposeful, and reality-oriented. Primordial cognition is concrete, irrational, and autistic. It is the thought of dreams and reveries. Primordial cognition is free-associative. This increases the probability of novel combinations of mental elements, which form the raw material for a work of art. This raw material must then be put into final form (e.g., be made to conform to current stylistic rules) in a more rational or conceptual state of mind. An extreme form of conceptual thought is deductive logic. It cannot possibly yield a creative idea, as the conclusion is implicit in the premises. If we know that
Chapter 15. Poetic History
all men are mortal and that Socrates is a man, then we can conclude that Socrates is mortal. However, we at least implicitly already knew this. We could compare the ideas of conceptual thought to atoms in a crystal. Each atom is firmly fixed in place, so that the probability of two remote atoms colliding and forming a new molecule is zero. When analyzed, creative ideas are always new combinations of old ideas that had been thought to be unrelated (Martindale 1995a). If we wish the atoms in a crystal to rearrange themselves, we can heat it. This will loosen the bonds amongst atoms and allow atoms distant from one another to collide and combine. Just so, regressing to a primordial state of mind makes our thinking loose and associative. Remote ideas are more likely to combine to form new ideas. Of course, the combinations are not random. The atoms (ideas) that will combine must have an ‘elective affinity’ for one another.
7. Style change and stylistic elaboration Novel ideas could emerge in two ways from the inspiration-elaboration process: holding the amount of elaboration constant, deeper regression toward primordial cognition should lead to more free-associative thought and thus increase the probability of new ideas. To produce a novel idea, one must regress to a primordial level. To produce an even more novel idea, one must regress to an even more primordial mode of thinking. Holding amount of regression constant, decreasing the degree of elaboration can lead to statements that are original by virtue or being previously disallowed or “nonsyntactic” to varying degrees. Because increasing the novelty of utterances by decreasing level of elaboration is more drastic than increasing novelty by increasing depth of regression during inspiration, poets seem to favor the method of increasing depth of regression, if possible, rather than the method of decreasing level of elaboration. If possible, successive poets should engage in deeper and deeper regression while maintaining the same level of elaboration. Each successive poet must regress further in search of usable combinations of words not already used by his or her predecessors. We should expect the increasing remoteness of similes and metaphors to be accompanied by content indicating the increasingly deeper regression toward primordial cognition required to produce them. Another way of putting this is that poets tend to work within styles. We may loosely define a poetic style as a lexicon of words that may be used in poetry and a set of rules governing how these words may be combined. The early poets writing
Colin Martindale
within a style will discover the obvious word combinations and use them. Once used, these combinations cannot be used again. Thus, later poets will need to regress further in order to find novel combinations of words that have not yet been used. Eventually, a turning point will be reached. At that time, further increases in novelty are easier to attain by decreasing level of elaboration – i.e., by loosening the stylistic rules governing the production of poetry. This corresponds to a period of major stylistic change. Hypothetically, stylistic change allows poets to return to word combinations composed of relatively close associates. This is accomplished either by changes in the poetic lexicon such that entirely new words are dealt with or by loosening the stringency of poetic rules so that previously forbidden word combinations are allowed. There should be a partial return from primordial toward conceptual cognition during periods of stylistic change: because the rules have been changed, deep regression is not needed to produce novel ideas. Once such stylistic change has occurred, the process of increasing regression would be expected to begin again. Empirically, it seems that poets first add words to and drop words from the poetic lexicon (Miles 1964). Only then do they give up and change the rules of the game. Early Romantic poetry provides an example of stylistic change involving a change in the lexicon. Once poets had said all they could think of about high and mighty things, they simply stopped writing about them and wrote about common people. Wordsworth’s poem, ‘The Leech Gatherer’ is a nice example of this tactic. The leech gatherer did not need to say anything interesting and did not do so. To attain novelty and elicit attention, Wordsworth merely had to introduce such a character into the poetic lexicon. There is not much to say about leech gatherers, so they made only a brief appearance in the history of poetry. A clear example of a stylistic change in rules can be found in the history of modern French poetry. Until 1900, French poets accepted the common-sense stylistic rule that the word “like” had to join like words. Thus, if a poet wanted to compose a simile, “A is like B,” then “A” and “B” had in fact to be alike in at least some arcane way. By the end of the nineteenth century, a lot of French poets had written a lot of poetry. It had become very difficult to comply with the stylistic rule without repeating what someone else had already said. Around 1900, this rule was quite explicitly abrogated (see Martindale 1975). It became acceptable poetic practice to combine unlike words with the word “like.” Thus, Eluard’s surreal image, “the earth is blue like an orange,” was perfectly good poetry. Surreal images tend to be composed of close word associates such as “blue” and “orange.” No great regression is needed think of “orange” given the word “blue”. I have elsewhere presented quantitative evidence that successive nineteenth-century French poets did in fact regress more and more up to about 1900, when the process was reversed
Chapter 15. Poetic History
and depth of regression decreased, presumably because of the loosened stylistic rules (Martindale 1975, 1990). If the theory is valid, three predictions can be made about any series of literary products produced within a given tradition: 1) indices measuring collative properties such as novelty, complexity, and variability should increase monotonically across time; 2) there should be cycles of increasing and decreasing density of words indicative of primordial cognition; 3) periods when primordial content decreases should show evidence of stylistic change.
8. A quantitative history of British poetry A number of studies have been conducted to test the evolutionary theory outlined above. Those concerning literature include investigations of nineteenth- and twentieth-century French poetry, fourteenth- through twentieth-century British poetry, seventeenth-century English metaphysical and non-metaphysical poetry, eighteenth- through twentieth-century American poetry, twentieth-century American popular music lyrics, and an experimental simulation of literary history. These studies are described in detail in Martindale (1990), as are studies of the history of painting, architecture, music, and science. Below, I outline the results for the study of British poetry. The epoch from 1290 to 1949 was divided into 33 successive 20-year periods. For each of these periods, the poets born during the period were ranked on the basis of number of pages devoted to them in the relevant Oxford anthology of English verse. For the last 20 periods – 1550 to 1949 – this caused no difficulties. The seven poets assigned the most pages were included in the sample. For the earlier periods, it was not always possible to find seven poets because of uncertainties about birthdates and because very little poetry from the earliest periods has survived. The sample consists of 170 poets. Once poets had been selected, the most complete and recent edition of their poetic works was obtained. Fifty random samples were taken by drawing 50 page numbers from a table of random numbers. The first eight lines were counted off and the sample for each page was terminated at the first phrase delimiter (e.g., period, semicolon, question mark) at or after the end of the eighth line. The mean number of words per poet was about 3,000, and the mean number of phrases per poet was about 200. Old spellings were consistently modernized to facilitate dictionary look-ups. Such modernization was confined to minor spelling changes. I end up with a sample totaling 521,566 words. Because of the large amount of text analyzed, computerized content analysis was employed. To attempt to test the theory using a traditional humanistic or
Colin Martindale
qualitative approach would have been impossible. The task of reading the works of 170 poets and deciding whether impact value increased in a monotonic fashion across time completely exceeds the capacities of human memory. The first question of interest concerns the prediction that the impact value of poetry has increased over time. I constructed a Composite Variability Index to measure the collative properties of texts. The goal was to create an index of the degree of complexity, surprisingness, incongruity, ambiguity, and variability of texts. In creating the index, several steps were involved. First, non-redundant measures with face validity were selected. Then, because many of them are spuriously related to the number of words or phrases in a text, scores with the effects of number of words and number of phrases statistically removed were computed. Finally, a Composite Variability Index was created by adding together the variables in standard score form (to give each equal weighting). The index is composed of measures such as the hapax legomena percentage (percentage of words occurring only once in a document), mean word length, the coefficient of variation of word frequency, of word length, and of phrase length. Deviation from the expected mixture of variation in word length or phrase length leads to a text that is more unpredictable. The Composite Variability Index is for the most part a measure of unpredictability or entropy. A text that is unpredictable should be surprising. The more unpredictable a poem is, the less certain we are what the poet is going to say next. The Composite Variability Index gets at unpredictability on a very basic linguistic level. The Composite Variability Index varies across periods in a highly significant way. Readers interested in the specific statistical tests used in determining this and other trends mentioned below are referred to Martindale (1990). Differences among the periods are much greater than differences within the periods. As predicted, these differences are due to a monotonic uptrend over time. The best fitting trend line is an exponentially increasing equation. The equation tells us that the rate of change in impact value has accelerated across time. Impact value has been accelerating according to the equation since before Chaucer. It is speeding up in its rate of change, but this has always been the case rather than being a modern phenomenon. To test the predictions concerning trends in primordial content, we need a measure of the latter. The Regressive Imagery Dictionary (Martindale 1975) contains 2,900 words assigned to 36 categories. Each word is assigned to only one category. The basic categories are summed to yield two summary categories that measure primordial and conceptual content. The primordial content categories are grouped into subdivisions of Drives, Sensations, Perceptual Disinhibition, Regressive Cognition, and Icarian Imagery. Each of these has been suggested by the
Chapter 15. Poetic History
various theorists mentioned above as being important in primordial cognition. The categories measuring conceptual content have, likewise been used by theorists in describing this type of thought. The dictionary in English, French, German, Latin, Russian, Swedish, and Portuguese versions along with Wordstat, a program that applies it to texts, is available from Provalis Research: http://www.provalisresearch. com/wordstat/RID.html. (The dictionaries are in the public domain). Martindale (1990) presents evidence concerning the rationale and validity of the coding scheme. A number of studies support the construct validity of this scoring scheme. No attempt to find evidence for the construct validity of the dictionary has ever failed. Theoretically, more primordial content should be found in the verbal productions of children (Werner 1948), primitive people (Werner 1948), individuals exhibiting symptoms of psychopathology (Freud 1900), people under the influence of consciousness-altering drugs, and of hypnosis (Fromm 1978), and people exhibiting a lot of right- as compared with left-hemisphere cortical activation (Hoppe 1977). The Regressive Imagery Dictionary has yielded results in conformity with each of these predictions in a series of 12 studies (see Martindale 1990, for descriptions and the original citations). Further evidence has come from seven studies derived from Jungian theory concerning trends within the content of literary narratives (see Martindale 1990; Martindale & West 2002) and of a study of co-occurrence of primordial content with Jungian archetypal figures (Martindale & Martindale 2007). Thus, the Regressive Imagery Dictionary does seem to yield a valid index of primordial or dedifferentiated thought in a variety of contexts in which the measure varies as is theoretically expected. In order to obtain a general measure of primordial content, the five primordial content categories are added together and the conceptual content category is subtracted from this sum. When the Regressive Imagery Dictionary was applied to the series of texts, primordial content oscillated across time in a manner that could not be attributed to chance. It rose while styles agreed upon by literary critics were in effect and fell as a new style was introduced. For example, it rose while the metaphysical style was dominant and began declining as this style was gradually replaced by the neo-classical style. Once the neo-classical style was firmly established, it began to rise again. In the statistical study of British poetry, there were enough epochs to measure the number of words added, dropped, and retained from the prior epoch. This was a difficult task, but the results were very clear. There were oscillations in number of words added and dropped (presumably measuring stylistic change). These were out of phase with the oscillations in primordial content. That is, when primordial content was rising, the number of words added and dropped declined; when primordial content was falling, the number of words added and dropped increased. This suggests that when a given style is in effect, rather few
Colin Martindale
words are added or dropped. As a new style is introduced, the number of words added and dropped increases. A given style stays in effect for several decades with impact value mainly being determined by ideas generated by increased novelty due to primordial cognition. Once the old style is exhausted, words are added and dropped at an increasing rate, and the index of primordial content declines, because the new words allow novelty to increase with less rather than more primordial cognition. If the words themselves are new, it is not necessary to combine them in new ways.
9. Range of the theory One may object that everything I have said is true, but that I have merely described several trivial trends that account for very little of poetic or artistic history. In the case of British poetry, I was able to compute statistics that show that almost half of the history of this poetic tradition can be explained by the theory about novelty and primordial content. This was done by computing a 170 × 170 similarity matrix showing the similarity of each poet to each other poet, doing a multidimensional scaling of the matrix to get rid of trivial dimensions, and then seeing how much of the variation in the multidimensional space could be accounted for by the theoretical variables. I am told that a recurrent nightmare of humanistic scholars is that someone like me will feed all the poetry through a computer, come up with an equation that explains everything, and put them out of business. They should not be concerned. The history of British poetry may be described by a complex trajectory through a five-dimensional space. A rule of thumb is that a phenomenon described by N dimensions requires N equations to describe it. In the case of British poetry, I have only approximated two of these equations and am not at present actively searching for the other three. As for the possibility that the trends that were found reflect trends in external society, this seems definitely not to be the case. I collected hundreds of series of social indicators ranging from the price of bread in London and average wages through rate of illegitimate births and intensity of social disturbances to net wind direction in England (a measure not only of the wild west wind but also of average temperature). After detrending where necessary, I correlated them with the theoretical variables as well as with categories from Stone’s (1966) Harvard III Psychosocial Dictionary, which categorizes virtually all words in a text into a number of general categories. In both cases, far less correlations than would be expected by chance were found. Nothing I could measure in British poetry is related to anything that anyone has been able to measure in the larger society. British poetry has evolved in a virtual social vacuum.
Chapter 15. Poetic History
References Berlyne, D.E. 1965. Structure and Direction in Thinking. New York NY: Wiley. Berlyne, D.E. 1971. Aesthetics and Psychobiology. New York NY: Appleton-Century-Crofts. Bogen, J.E. 1969. The other side of the brain: An apositional mind. Bulletin of the Los Angeles Neurological Societies 34: 135–162. Bruner, J. 1991. The narrative construction of reality. Critical Inquiry 18: 1–21. Cassirer, E. 1925/1955. The Philosophy of Symbolic Forms. New Haven CT: Yale University Press. Darwin, C. 1859. On the Origin of Species. London: Watts and Co. Darwin, C. 1871. The Descent of Man and Selection in Relation to Sex. New York NY: D. Appleton. Darwin, E. 1789/1791. The Botanical Garden. London: J. Johnson. Freud, S. 1900/1994. The Interpretation of Dreams. New York NY: Barnes and Noble. Fromm, E. 1978. Primary and secondary process in waking and in altered states of consciousness. Journal of Altered States of Consciousness 4: 115–128. Goldstein, K. 1939. The Organism. Boston MA: Beacon Press. Hobbes, T. 1642/1958. Leviathan. New York NY: Liberal Arts Press. Hoppe, K. 1977. Split brains and psychoanalysis. Psychoanalytic Quarterly 46: 220–244. Jung, C.G. 1935/1972. The Relations between the Ego and the Unconscious. In The Collected Works of C.G. Jung, Vol. 7, H. Read, M. Fordham & G. Adler (eds), 123–241. Princeton NJ: Princeton University Press. Klinger, E. 1971. Structure and Functions of Fantasy. New York NY: Wiley. Kris, E. 1952. Psychoanalytic Explorations in Art. New York NY: International Universities Press. Lévy-Bruhl, L. 1910. How Natives Think. New York NY: Washington Square. Martindale, A.E. & Martindale, C. 2007. Portrayal of women and Jungian anima figures in literature: Quantitative content analytic studies. In Aesthetics and Innovation, L. Dorfman, C. Martindale & V. Petrov (eds), 212–233. Cambridge: Cambridge Scholars Press. Martindale, C. 1975. Romantic Progression: The Psychology of Literary History. Washington DC: Hemisphere. Martindale, C. 1981. Cognition and Consciousness. Homewood IL: Dorsey. Martindale, C. 1990. The Clockwork Muse: The Predictability of Artistic Change. New York NY: Basic Books. Martindale, C. 1995a. Creativity and connectionism. In The Creative Cognition Approach, S. Smith, T. Ward & R. Finke (eds), 249–268. Cambridge, MA: The MIT Press. Martindale, C. 1995b. The second law of thermodynamics is the first law of art history. In Informational Approach and Art Studies, I. Gordova, V. Petrov & Y.N. Rags (eds), 44–52. Moscow: Krasnodar. Martindale, C. 2007a. A neural-network theory of beauty. In Evolutionary and Neurocognitive Approaches to Aesthetics, Creativity, and the Arts, C. Martindale, P. Locher & V. Petrov (eds), 182–193. Amityville NY: Baywood. Martindale, C. 2007b. Recent trends in the psychology of aesthetics, art, and creativity. Empirical Studies of the Arts 25: 121–141. Martindale, C. 2007c. Sociocultural oscillations and their analogies with physical waves. In Aesthetics and Innovation, L. Dorfman, C. Martindale & V. Petrov (eds), 332–369. Cambridge: Cambridge Scholars Press. Martindale, C., Moore, K. & Borkum, J. 1990. Aesthetic preference: Anomalous findings for Berlyne’s psychobiological theory. American Journal of Psychology 103: 53–80.
Colin Martindale Martindale, C. & West, A. 2002. Quantitative hermeneutics: Inferring the meanings of narratives from trends in content. In Thematics: Interdisciplinary Studies, M. Louwerse & W. van Peer (eds), 377–396. Amsterdam: John Benjamins. Maslow, A.H. 1957. Two types of cognition and their integration. General Semantics Bulletin 20: 17–22. McKellar, P. 1957. Imagination and Thinking. New York NY: Basic Books. Miles, J. 1964. Eras and Modes in English Poetry. Berkeley CA: University of California Press. Mukařovský, J. 1940/1976. On Poetic Language. Lisse: Peter de Ridder. Neisser, U. 1967. Cognitive Psychology. New York NY: Appleton-Century-Crofts. Nietzsche, F. 1872/1994. The Birth of Tragedy out of the Spirit of Music. New York NY: Penguin. Petrov, V.M. 2003. Cyclic cultural evolution against the background of long-range progressive trends: Information approach. Journal of Cultural and Evolutionary Psychology 1: 85–107. Piaget, J. 1936/1952. The Origins of Intelligence in Children. New York NY: International Universities Press. Shklovsky, V. 1919/1969. Der zusammenhang zwischen den verfahren der Subjektfugung und den allgemeinen Stilverfahren. In Texte der Russichen Formalisten, J. Štriedter (ed.), 37–42, Munich: Fink. Stone, P.J. 1966. The General Inquirer: A Computer Approach to Content Analysis. Cambridge MA: The MIT Press. Taine, H. 1863. Histoire de la Litterature Anglaise. Paris: Hachette. Tynjanov, J. 1929/1967. Archaisten und Neuerer. Munich: Fink. Vico, G. 1730/1744/1948. The New Science of Giambattista Vico. Ithaca NY: Cornell University Press. Werner, H. 1948. Comparative Psychology of Mental Development. New York NY: International Universities Press. Wundt, W. 1896. Lectures on Human and Animal Psychology. New York NY: Macmillan. Zajonc, R.B. 1968. Attitudinal effects of mere exposure. Journal of Personality and Social Psychology Monograph Supplement 9(2, pt. 1): 1–27.
chapter 16
Consolidating empirical method in data-assisted stylistics Towards a corpus-attested glossary of literary terms Bill Louw Method (Gk. Meta + hodos) means an ‘after-path’. Radical revisions of methodology follow momentous paradigm-shifts within scientific theories. Hence linguistic-stylistics developed analogue collocation into its digital counterpart, especially through the discovery of semantic prosodies (Sinclair 2004b; Louw 1993). This led to the recognition (Louw 1991; 2000; 2007d) that all literary devices have a corpus-accessible feature in common: relexicalisation. Delexicalisation arose out of developments in lexicography.
Sinclair refers to the two terms as forming a continuum (Sinclair 2004a: 198fn18). This continuum is marked (Enkvist 1973), unlike Hoey’s (2005) purported, but psychologist priming. He omits Firth’s (1957) pre-condition that collocation is abstracted from syntax and that collocative (relexicalising) power falls off within four words on either side of a node. This paper explores the consequences for science and glossaries of literary terms of collocation as instrumentation for meaning. Keywords: Data-assisted stylistics, corpus-attested glossary, semantic prosody, relexicalization, collocation, lexicography
1. Introduction This paper seeks to consolidate the scientific use of empirical methods in the examination of literary texts by: (1) demonstrating the manner in which those methods arise out of research in corpus stylistics and (2) preventing their hybridization in practice through the creation of intuitively-derived concepts or procedures. This objective will be difficult to set in place unless and until a
Bill Louw
corpus-attested glossary of literary terms has been derived from the research. All entries in such a glossary will need to be fully testable by means of scientific criteria such as replicability. Such tests are consistent with the objectives and practices supported by this commemorative volume. One of the proofs that digital collocation is nascent instrumentation for language is to be found in the fact that all literary devices may be shown to share a common collocational trait: relexicalization (Louw 2007b; Louw, forthcoming). Relexicalization can only be defined in relation to its continuum-based counterpart delexicalization. When Sinclair et al. (1987; 1993) was faced with the difficulty of describing, in a dictionary, the large number of meanings (in excess of 80!) occupied by highly frequent words such as take, he referred to them as delexical or ‘washed out’. For example, the phrase take a look may be replaced by have a look or simply look. Furthermore, the delexical forms were found to occupy positions that were more frequent than their fully lexicalised counterparts: the most frequent meaning of the term see, occupies the meaning ‘understand’ rather than any meaning that relates to ‘vision’. Hence, Sinclair uses the delexical form as the term’s first meaning in his dictionaries. Relexicalization is a collocational phenomenon that takes place when collocates occur within 4 words of a node: I see with my own eyes. Relexicalized forms are marked in Enkvist’s (1973) sense of the term. Relexicalization is brought about in two ways only: (1) by the appearance of two or more collocates within the 9 word window of collocative power (Sinclair 1991: 175; 2004a: 195 fn18) and (2) the uncontextualized use of single words as citation forms (Philip 2003: 235). The phenomenon which was to be called relexicalization caused Frege (1884) to caution logicians and philosophers never to countenance the use of single word examples or citation forms, for the reason that his distinction between Sinn (sense) and Bedeutung (reference) could only be established by the presence of what we now know to be a crucial collocate, acting as Argument to a Function. This is exemplified in his celebrated example concerning reference to the planet Venus as the morning star or the evening star. Burge (2005) sums up what Frege saw as the impact of single words on truth-value “… insofar as they have truth-value, the single words mask an implicit structure, involving some sort of implicit predicational attribution.’’ (Burge 2005: 133) (emphasis added) Ironically, Burge’s summary encapsulates the main concerns of digital collocation: to find collocation’s implicit non-syntactic structures of meaning that rise above and are abstracted from syntax (Firth, in Palmer, [1957] 1968: 197) and are hence, free from predicational attribution. Collocation is capable of re-assigning mere grammatical categories as part of the process of symbolism.
Chapter 16. Consolidating empirical method in data-assisted stylistics
By way of contrast, Hoey’s (2005) averment that ‘words’ are ‘primed’, in some unexplained and psychological way, forces his work to remain unproven and probably unprovable (Louw 2007c). The early work of Roman Jakobson on the subject of the poetic function (in Sebeok 1960; and in Weber 1996) pointed, albeit weakly, in the direction of relexicalization. He had initially described as the ‘set toward the message’, a characteristic that determines how poetic language focuses upon itself. Cook (2000: 192) considers this process to have been only weakly described. However, even the early version of the functions were better marked than Hoey’s priming, because Jakobson’s system of functional differentiation within his six functions can be tested for markedness using the commutation test proposed by Louw (1991: 156). Thus, even during the analogue period of language study, the poetic function established conclusively that there can be nothing inadvertent about the intention behind poetic language. The Russian Formalists had always asserted that poetic language would be distinct from what might be best referred to as everyday language. Formalists often demonstrated how the banal act of language involved in, say, purchasing apples would always be different from what they referred to as self-valuable reference. This involved the selective vision of the poet, regarding with the eye of Shakespeare’s Theseus, an apple, starting perhaps with its stalk, shape, color, smoothness or texture and always with some hint ‘… of Man’s first disobedience, and the fruit of that forbidden tree’ in order to accord to symbolism (Gk. ‘that which is thrown together’: syn + ballein), ‘… a local habitation and a name’. Today, however, digital collocation renders falsifiable Jakobson’s poetic function (Louw 2007b: 150), which, even in its most elaborated form is now 50 years old (Weber 1996: 33n). The poetic function was falsifiable because it remained hobbled by the fact that it operates syntactically, whereas collocation operates at a level described by Firth, as we have seen, as ‘abstracted from syntax’. Firth locates collocation just below his own language levels for context of culture and context of situation (Louw 2007b: 154). In 1957, Firth provided John Sinclair with the impetus for conducting his research into digital collocation that led to the production of the osti Report (Krishnamurthy 2004). Firth’s now famous observation ran as follows: “Meaning by collocation is an abstraction at the syntagmatic level and is not directly concerned with the conceptual or ideal approach to the meaning of words. One of the meanings of night is its collocability with dark […]” (in Palmer, [1957]1968: 197). Digital collocation had been established by Sinclair as early as the late 1960s, but sadly, even the British Library ‘lost’ its only copy of the original osti Report and attempts to publish it more than 40 years later were not free from apparent interference (Sinclair, personal communication).
Bill Louw
2. Literary devices and the delexical-relexical continuum There has always been the possibility that the delexical, or what John Sinclair frequently referred to verbally, but never in writing, as the ‘blue jeans’ or ‘washed out’ meanings of everyday events might be incorporated into poetry. Sinclair used to describe it to his students in this way (personal communication): “The more one washes blue jeans, the more they fade. The more one uses an English word or expression, the more its meanings are washed out progressively into different meanings.” The mechanism for marking this phenomenon in Enkvist’s terms was to remain unrecognized until the advent of the information age: it is a collocational phenomenon and corpus linguists refer to it as relexicalization. There is a sense in which the everyday references which we find in poetry become more deeply associated with their origins through the agency of the very genre of modern poetry. Within this genre they remain frozen for all time in a more fully lexicalized form than any routine, everyday event might ever dictate. Such relexicalization endures as a powerful ingredient of symbolism. Much ink has been spilled on the entirely putative subject of connotation. Collocation is, by comparison, recoverable by computational means and its status is evidential. A good example of what might be called monumental or genre-evoked relexicalization is to be found in the bravado of Fleur Adcock’s persona in her poem ‘Against Coupling’. The poem’s persona rebels against all the fake posturing involved in human sexuality and insists that all that is needed is five minutes alone and that they are sufficient to fill … that gap between the Sunday papers and lunch. (in Morrison & Motion 1982)
Sexual needs are brushed aside and the gap to be filled is re-cast as intellectual or culinary, notwithstanding the fact that the poet’s own surname may be suggestive of an absence (Barthes 1967; Macherey 1966). However, the corpus is capable of offering a deeper insight into life’s spiritual and human vacuum when the terms gap and life are co-selected. Five minutes’ solitude is no match for these. MicroConcord search SW: gap CW: life 80 characters per entry Sort : 1R/SW unshifted. 1 aps its the span which represents the 2 do well, but there was an increasing 3 that term, but he wanted to close the 4 nce of payments deficit; the widening 5 as come to the awful realisation of a 6 my sister-in-law Kate has filled the 7 uncritical acceptance of you fills a 8 ean-Claude's life. Her death opened a 9 uccessful actor. He will leave a huge 10 id that his mother's death had left a 11 go of Lord Zuckerman will leave a sad
gap gap gap gap gap gap gap gap gap gap gap
between life and death? Athelstan rema between my home life and my aspiration between the gospel and life. In Pacem between conditions of life and work in in her waning life which will never be in my life by producing a baby brother in your own life. Never underestimate in his existence which threatened cont in the theatrical life of Broadway; fo in his life which nothing could fill. in public life in Britain and elsewher
Chapter 16. Consolidating empirical method in data-assisted stylistics
12 er a bitter struggle. They had left a 13 hould ask yourself why there's such a 14 d it- Since she's gone there's been a 15 'm not the woman you need to fill the 16 ing for his life out of the narrowing
gap in the life of the city. She contraste gap in your life. If you're seeking a way gap- it made my life better, because — she gap Lotta left in your life! Why gap.
He made it. Fucking bastard!
(Source: The British National Corpus)
Malinowski’s (1935) routine associations of gardening or fishing could not fail to become more profound symbolically if they were ever to be set within the boundaries of a poem. The very simplicity of their primitivism, in Fortunati’s (forthcoming) sense, relexicalizes them and breathes life into their symbolism. At the beginning of St. Mark’s Gospel we witness the irony, akin that of the Sartor Resartus, of fishers being fished by the Son of Man – snared within the toils of a faith whose moment, John the Baptist declared, had arrived. It is in these expressions that the collocates of fishing as a Wittgensteinian (1922) Sachverhalt or ‘state of affairs’ are suddenly altered and propelled into a new symbolism by means of a hitherto unknown collocate as a ‘fish-surrogate’: man himself is fished and the disciples are to become ‘fishers of men’. Is this device more profound than the apparent paronomasia of Philip Larkin (1988) in his description of home in the poem of the same name? A glorious shot at how things ought to be Long fallen wide …
What looks like a washed out Lakoffian-Johnsonian (1980; 2003) ‘metaphor we live by’ at the beginning of the sentence is imbued, by the time we reach its end, with the dark blue of new blue jeans, as though they had never known the routine washdays of married life and of home. It is relexicalized: the conflicts of marriage are terrifyingly more real than the mere delexical attempt involved in ‘having a shot at’ something. Having a shot at marriage involves real priming and dry powder. 3. Collocation and critical methodology The term method is derived from the Greek meta (after) + hodos (path). Its etymology is suggestive of change: method is the ‘after-path’ that follows momentous change of a type that frequently involves a paradigm-shift. The remainder of this paper will examine the way in which methodology in stylistics is likely to alter as a result of the digital turn away from its analogue past. The terms digital turn and empirical need to be examined both separately and in relation to one another. The terms digital and analogue are derived by extension from the way in which they are used in science to refer to technological advances in computers. A computer that operates on data in the form of digits, rather than
Bill Louw
the physical quantities used in analogue computers (employing cogs, gears and levers), is essential for dealing with the task of stylistics as a form of data-assisted reading. The reason for this is that a corpus acts as a sample of the whole of the language of the target text. Any particular instance of language from a literary text under examination and for discussion needs to be supported by the empirical power of a digital computer to produce more contexts for that linguistic feature from the whole language than mere intuition or introspection might afford. Digital computers are capable of dealing automatically with a very wide range of problems at very high speeds. This takes us to the way in which the term empirical has altered within critical theory in order to warrant the use of the expression digital turn. For example, Belsey (1980: 7) states that common sense in literary criticism proposes a form of humanism based on an empiricist-idealist interpretation of the world. Her use of the term empiricist implies no more than reliance upon experience. Philosophers have always used the term in this way. However, for the scientist, the term always refers to the results of experiment and observation only. It is towards this meaning and with the objectives of this timely commemorative volume in mind that the digital turn is being made. This paradigm-shift is already underway. Once completed, it is likely to bring an entirely new order into the methodology of literary stylistics and criticism. This new order is likely to alter the procedural steps of traditional literary criticism and stylistics. Note how the approach of Firth and Malinowski differs from those of Abrams and Jakobson in Table 1. Instead of beginning with speech or writing as phonic or graphic data, they begin with the text embedded in its context of culture and situation (Malinowski stated that translation becomes impossible unless this is done). The approach of Abrams owes more to psychologism, but the approach of Firth and Malinowski is to be preferred because of their commitment to banishing concepts and the ‘ideas’ meaning of words. Collocation ‘pictures’ these contexts of culture and situation in Wittgenstein’s (1922) sense and as part of his Picture Theory of Meaning which the Tractatus sets out in detail. Table 1. Typical approximate approaches to stylistics and criticism Direction of process is downward in each column Jakobson (1960)
Firth & Malinowski (1935)
Abrams (1957; 1970)
phonetic phonological morphological grammatical syntactic semantic discoursal
context of culture context of situation stylistics collocation lexicography grammar and syntax phonaesthetics
thought-form rise and fall of emotions figures of speech imagery diction sound
Chapter 16. Consolidating empirical method in data-assisted stylistics
The breakthrough in the area of digital collocation that was to facilitate the alteration of the methodology set out above has come about fairly gradually. Attempts were made to utilize digital collocation in literary stylistics as early as a paper by Louw (1989) delivered in 1987: the year in which the first edition of the Cobuild English Language Dictionary was produced by John Sinclair. However, the approach could not be consolidated until technological advances had taken place. For example, in 1987 the corpus amounted to a mere 21 million words of running text and computers were incapable of carrying out procedures such as coselection and the creation of N-grams. The Bank of English is currently operating at a total of half a billion words of running text. Both of the procedures mentioned are crucial to the computational construction of literary worlds. This is especially advantageous for second language learners. The ‘states of affairs’ inherent in literary devices will be fictional and hence modified versions of everyday events. Second language learners practicing corpus stylistics will need only two collocates in order to retrieve a particular state of affairs from the corpus using co-selection. As the state of affairs emerges from the corpus, it will carry with it all of its other collocates. The practitioner will then be in a position to determine exactly which collocates are present and which collocates have been omitted by the literary author. These insights are crucial to both language learning and to literary appreciation. The much-vaunted integration of language and literature has finally come of age. Advances of this theoretical approach in the digital domain continue to be made in spite of censorship, particularly of (1) any reference to the role of Michael Halliday (in Bazell 1966: 156) in reining back Firthian collocation into syntax; (2) the role of semantic prosody in the falsification of linguistic theories (Louw 2007a), and (3) the connection between collocation as instrumentation for language, philosophy and science (Louw 2003). The most serious censorship was carried out by Michael Hoey upon two of this author’s chapters that were to have appeared in the much delayed Approaches to Corpus Stylistics, a volume that was to have been published by Routledge. His apparent motive was to replace scientifically respectable semantic prosody with his own almost entirely intuitive notion of priming. Whitsitt (2005) appears to have had a similar motive. His entirely non-computational paper inexplicably appeared in flagrant contravention of the published criteria for the acceptance of corpus-based research by the International Journal of Corpus Linguistics. However, it was John Sinclair himself who brought about the realization that the context of culture and the context of situation are now accessible within corpora of natural language by empirical means and in fulfillment of Firth’s vision for linguistic and stylistic fieldwork. Sinclair’s pamphlet entitled Phrasebite is set out in full below.
Bill Louw
“When she was – Phrasebite© John Sinclair, 2006. 1. The first grammatical collocate of when is she 2. The first grammatical collocate of when she is was
3. The vocabulary collocates of when she was are hair-raising. On the first page: diagnosed, pregnant, divorced, raped, assaulted, attacked The diagnoses are not good, the pregnancies are all problematic. 4. Select one that looks neutral: approached 5. Look at the concordance, first page. 6. Nos 1, 4, 5, 8,10 are of unpleasant physical attacks 7. Nos 2, 3, 6, 7, 9 are of excellent opportunities 8. How can you tell the difference? 9. the nasties are all of people out and about, while the nice ones are of people working somewhere. 10. Get wider cotext and look at verb tenses in front of citation. 11. In all the nasties the verb is past progressive, setting a foreground for the approach. 12. In the nice ones, the verb is non-progressive, either simple past or past-in-past. Data for para 4 above. (1) walking in Burnfield Road, Mansewood, when she was approached by a man who grabbed her bag (2) teamed up with her mother in business when she was approached by Neiman Marcus, the department store (3) resolved itself after a few months, when she was approached by Breege Keenan, a nun who (4) Bridge Road close to the Causeway Hospital when she was approached by three men who attacked her (5) Drive, off Saughton Mains Street, when she was approached by a man. He began talking the original (6) film of The Stepford Wives when she was approached by producer Scott Rudin to star as (7) bony. Kidd was just 15 when she was approached to be a model. Posing on (8) near her home with an 11–year-old friend when she was approached by the fiend. The man (9) finished a storming set of jazz standards when she was approached by SIR SEAN CONNERY. And she (10) on Douglas Street in Cork city centre when she was approached by the pervert. The man persuaded (from: Phrasebite, John Sinclair 2006)
The power of this publication, coming as it did so close to Sinclair’s death, is to be found in the detail of his method. By beginning with a single word, she, from the whole of the Bank of English, Sinclair simply requests the most
Chapter 16. Consolidating empirical method in data-assisted stylistics
frequent collocate from the Bank of English (approximately 500 million words of running text). The computer provides it: when. The results are then merged: when + she. A new search is initiated for the most frequent collocate of this twoword phrase. The computer provides it: was. The concordances are scrutinized and cultural insights are gathered. This may appear to be a fairly unremarkable exercise, but its methodology is highly significant. Notice that no structural criteria are invoked by the investigator. Sinclair is not asking for sentences. Collocation has no interest in sentence boundaries. It is abstracted from syntax and operates above structure as we have known it for the past 3500 years. The significance of this finding is that Sinclair has opened for empirical study the context of culture and the context of situation. As a result, it becomes easy to determine the precise degree of under-provision or over-provision (Louw 2000) of those contexts that particular authors have chosen to offer the reader.
4. The path of methodology after the ‘digital turn’ in stylistics The impact of Sinclair’s Phrasebite is likely to alter radically the way in which stylistic methodology is ordered and perceived for use after the digital turn. This is likely to result in two main types of alteration to existing methodology: (1) the order in which the stylistic investigator proceeds and his/her commitment to scientific methods, and, (2) the re-writing of all glossaries of literary terms in order to make their definitions scientific rather than simply intuitive. We shall examine each of these in turn. 4.1 Re-ordering stylistic investigation There was a period during which stylistics slavishly followed the same order and procedures as linguistic fieldwork. A good example was the work of Jakobson. He was adamant that stylistic inquiry should deal primarily with stylistic concerns so as not to run the risk of replicating all of one’s primary data (Culler 1975: 73). However, because the work of stylistics was forced to run parallel to linguistic inquiry, stylistics of this kind began to suffer from all of the myopia of linguistics itself and, in consequence, took very few steps to check the nature of what was being sought in the name of stylistic evidence. The preoccupation at the time with the notion of style as deviation did little to uncouple the study of style from that of the norm against which style was purportedly to be recognized. When Jakobson was writing about the poetic function, the delexical-relexical continuum was unknown for the reason that the Cobuild technologies of John Sinclair had not yet begun to disclose it. In a
Bill Louw
lengthy footnote in Trust the Text, Sinclair (2004a) comments upon the delexicalization-relexicalization continuum and its potential use for stylistic investigation (see also Louw 1991). W. E. Louw (personal communication) argues that ‘literal’ and ‘figurative’ are points close to the extremities of a continuum of delexicalization. Words can gradually lose their full lexical meaning, and become available for use in contexts where some of that full meaning would be inappropriate; this is the so-called figurative extension. Louw points out that a writer, especially a literary writer, must exercise vigilance so that the meaning of each word is interpreted at the intended point on the continuum. Such features as collocation are part of the control mechanism available to the writer. (emphasis added) (Sinclair 2004a, 198)
By opening up the contexts of situation and culture for direct empirical examination at the level of collocation, Sinclair discarded once and for all time stylisticians’ reliance upon structure into the future. It is now possible to begin stylistic investigation at the far end of what used to appear to be the telos of stylistic investigation: literary worlds. The direction of investigation set out under the heading Firth and Malinowski in Table 1 above has now been made fully possible because it has been automated by Sinclair through digital collocation, as demonstrated by his method in Phrasebite. Stylistics has freed itself from any sense of duty or compulsion to follow the levels and direction of linguistic field work. 4.2 Th e notions micro and macro as criteria for the ordering of investigation Method has always been a major interest of stylisticians, although the apparently jumbled nature of their work often suggests otherwise. The analogue tradition in stylistics has its own famous methodological milestones, such as the objectivesubjective criteria insisted upon by Riffaterre (1971), and the Fowler-Bateson debate. But the arrival of the corpus began to look remarkably like the answer to the verification principle of the logical positivists. The use of a corpus for the purpose of literary stylistics seemed likely to invite eclectic verification of any detail provided by the investigator (micro criteria) rather than supplant analogue methodology overnight with full automation (macro criteria). There was a further advantage: the corpus looked set to democratize criticism by banishing esoteric practices and attitudes replacing them with objectivity. 4.2.1 Micro tasking and collocation One major attraction of using digital collocation as the instrument for carrying out a data-assisted reading is that micro tasks can be accomplished very quickly and easily using a corpus of natural language in conjunction with the target
Chapter 16. Consolidating empirical method in data-assisted stylistics
literary texts in a machine-readable form. For example, the status of a keyword such as ‘common’ in the novel Great Expectations can be very easily checked by issuing a search command that caters for this micro-task alone and comprehensively, through the use of wild-card methods: *common* (Louw, 2007e: 93). A larger task may involve searching the ‘rise and fall of emotions’ in a single poem by Hopkins (Louw 2007e: 103). However, since the publication of Phrasebite by John Sinclair, investigators need no longer balk at undertaking macro tasks that may operate across a whole work of art and across all of contemporary society. It is the search for these features that begins to allocate new priorities for the entire undertaking of data-assisted readings using digital collocation and natural language corpora. Sinclair’s method in Phrasebite impels the investigator to respect frequency as a criterion for the prioritization of methodology. Close reading need no longer be dominated by any sense of duty to proceed from a detail in the target text, such as the claw, as part of a journey towards considering the lion, according to Spitzer’s (1948) famous maxim: ex ungue leonem (from the claw … the lion). The way is open for moving above syntax and dealing directly and empirically with worlds, both real and fictional (Louw, 2007a). 4.2.2 Macro tasking and the ordering of investigation At the very beginning of the Tractatus Logico-Philosophicus and as early as its second line (1.1), Wittgenstein (1922: 7) offers us a theory of worlds that simultaneously invites expansion into literary worlds and their automation by means of digital collocation. 1.1 The world is the totality of facts not of things.
The naïve analogue view of collocation once fancied that collocation would be concerned primarily with the ontology of things. We see evidence for this even in Firth’s taxonomy for the incorporation into language study of Malinowski’s context of situation. A: The relevant persons, personalities; B: The relevant objects…
(Firth 1957: 182)
However, it becomes immediately apparent once digital collocation is applied to Wittgensteinian facts or Sachverhalten, i.e., states of affairs, that these are capable of accomplishing the macro-task not only of sketching the literary world of a work of art, but also, and for the first time, of determining its symbolism (how it is ‘thrown + together’, Greek: syn + ballein). This is accomplished by examining not only the collocates that occur in the text, but also by examining those collocates which are empirically recoverable in natural language, but which are not actualized in the text by the author. Such collocates are paradoxically empirically present, but
Bill Louw
do not occur. Developments of this kind must begin to impinge upon the vision of literary theorists such as Barthes (1967) and Macherey (1966). This type of activity is indicative of the paradigm-shift involved in the transition from analogue to digital stylistics. The term digital stylistics is preferred to corpus stylistics for the reason that the term corpora often applies to published work that is based upon the use of unsuitable collections of text, such as (a) small homemade, annotated collections; (b) sub-corpora associated with particular genres or language varieties, or (c) corpora that are dated and manifestly too small to act as reliable samples of the whole language. The Brown Corpus and the lob corpus fall into the latter category. It is likely to exert a great deal of influence on the question of the ordering of stylistic investigation’s ‘after path’ now that the digital turn has been brought about (Louw 2008, forthcoming). It will determine the method of stylistics well into the future because it heralds the status of collocation as instrumentation or the control mechanism (Sinclair 2004a: 181) for meaning. In the same way that Sinclair’s Phrasebite operates in the context of culture and situation in the real world, so symbolism sketches the operation of the literary world. The collocates common to both actual and virtual worlds are material to the creation of the poem’s symbolism. The following example sets this out in detail. It formed part of a presentation by the author (Louw 2007d) at the acorn Corpus Symposium in honor of the late John Sinclair delivered at Aston University in May 2007. Kismet Opal fires in the Western sky (For that which is written must ever be), And a bullet comes droning, whining by, To the heart of a sentry close to me. For some go early, and some go late (a dying scream on the evening air) And who is there that believes in Fate As a soul goes out in the sunset flare? R.B. Marriott-Watson Killed in action, 1918
(Source: Up the Line to Death: The War Poets 1914–1918. An anthology arranged by Brian Gardner (1982). London: Methuen.) The forms that appear in italics in the poem are, upon data-assisted scrutiny, capable of segmenting, or to adapt Sinclair’s (1991: 132) term, ‘chunking’ the poem’s literary world by means of a shared delexical collocate such as fire. The term fire is common to several Wittgensteinian states of affairs. However, as these are brought into proximity with one another, they proceed to relexicalize one another within the nine word window of collocative power. This occurs across the entire fabric of the poem (Louw 1991). Once relexicalization begins in the poem, the
Chapter 16. Consolidating empirical method in data-assisted stylistics
nine-word window expands the ambit of its power to accommodate all of the instances that the process symbolism requires. This phenomenon is also dealt with elsewhere by this author as the ‘zombie effect’. This refers to the process by which dead metaphors twitch when they are relexicalized through the power of collocation (2007b: 160). Relexicalization, by means of shared collocates, sketches the poem’s literary world. This activity explains, for the first time in digital stylistics, the way in which symbolism operates in its ancient sense. Observe how the collocate fire, occurs separately in the four states of affairs (Sachverhalten): opals, Western skies, bullets, and flares. From these it expands semantically by means of the collocation’s power to relexicalization the shared collocate fire into a symbol. Note carefully, also, how collocates may appear in the collocation lists, but not in the poem. This startling phenomenon remained almost entirely unnoticed by literary critics during the analogue period of stylistics. Macherey (1966) referred unwittingly to the phenomenon as absence, just as we saw that the term life is the absent collocate of gap in the poem by Fleur Adcock earlier. However, the appearance of an absence as empirical evidence is an entirely corpus-based phenomenon and constitutes proof that there has been a paradigm shift in stylistics after the digital turn. For this reason, the collocation lists from the bnc for the literary world of the poem Kismet are set out below. In the collocation lists that follow, the number of citations obtained appears adjacent to the form at the top of the list. Notice how in some of the columns the most frequent fully lexicalized word is sometimes more frequent than the collocate that is shared by several states of affairs. Such instances have a bearing upon the tone of the literary criticism which these frequent forms begin to generate. For example, gybe emphasizes the fact that flare is associated with anger. Senseless loss of life during wars provokes anger. The injuries (wounds and head) caused by bullets outweigh the mere fact that they are fired. The bullet in the poem simply comes to do its work. The concordances from the bnc for Western sky deliver four lines of fiery descriptions. A larger corpus such as The Bank of English would improve upon this by providing greater detail.
. The author conducted a nine-hour workshop at the University of Granada, Spain entitled ‘Establishing the Digital Turn’ on the 50th anniversary of Roman Jakobson’s poetic function and dedicated to the memory of the late John Sinclair. The workshop was held between 14 and 18 January, 2008. A number of students at the University of Granada have agreed to take further, as part of their own research, a project to create and update a glossary of corpus-attested literary terms.
Bill Louw
COLLOCATION LISTS for the poem KISMET opal* (130)
western sky (16)
bullet (792)
the 43 the 16 the 160 P 33 sky 16 in 124 a 29 in 5 P 118 of 28 was 4 ← and 69 and 27 his 61 in 13 a 52 with 12 through 39 an 10 to 37 you 10 from 31 fire 9 ← him 26 it 22 into 22 wound 22 head 22 fired 21 kismet (10) believes in (233) P 5 of 3 new 3 H 3 Fate 2
←
the he and that P a who of
91 57 50 38 35 31 30 29
flare (358)
← ←
the a of P and up to in from that her into it with was at by this for after had which as off could when gybe another is sudden you one she fired
169 121 96 85 68 48 41 36 18 18 18 17 17 17 16 15 14 14 14 14 13 12 12 11 11 11 11 10 10 10 10 9 9 8
← ← ←
Chapter 16. Consolidating empirical method in data-assisted stylistics it in she but if one God
20 20 15 14 13 12 11
←
Note how all of those ‘states of affairs’ that share the collocate fire not only relexicalize the term throughout the work of art, but utilize the logical space (Wittgenstein, 1922: 17) occupied by these states of affairs in order to sketch the literary world and draft the textual literary criticism which pertains to it. As opals are rotated disclosing their fire, so their symbolism resembles a human life governed by fate or God. Logical space demands that the person who fires the bullet cannot be the same Person that fired the flare of human life. That Person is God: the unspoken logical link between the two parts of the literary world: fire and fate. 2.19 ‘Logical pictures can depict the world.’
(Wittgenstein 1922: 17)
5. Method is derived from the priorities revealed by research Corpus stylistics only began in earnest once natural language corpora had become large enough to represent a reliable sample of natural language in action. For example, it was once traditional for corpora to amount to exactly one million words of running text. The lob and Brown corpora followed this model. Even the first Cobuild corpus and its reserve corpora came in at 21 million words of running text. This greatly enhanced the potential of the corpus for ‘sampling’ the whole language in relation to providing a data-assisted reading of any single line of a literary text provided by the stylistician. Fortunately, this has never been seriously disputed, because the research and development of such corpora was lexicographic and therefore always had this type of sampling as its main criterion. The development of a corpus large enough to enable the automated creation of two editions of Cobuild English dictionaries was the brainchild of John Sinclair. The first computerconcordanced dictionary in the world emerged in 1978. The first paper in corpus stylistics (Louw 1989) was delivered at the British Council workshop on language and literature at St Hilda’s College Oxford in April 1987. From 1987 to the present day, the methodology for corpus stylistics has advanced only in relation to the major findings of the research, as the table below seeks to elucidate. The third column of the table is the ultimate objective of this paper, because whenever a paradigm shift takes place or is accomplished in a discipline, its terminology needs to be systematically and scientifically re-defined.
Bill Louw
The digital turn marks such a paradigm-shift because it marks the moment at which stylistics, by means of collocation, freed itself from being treated as a structural and syntactic phenomenon. The boundaries of science in this regard were not only of interest to Firth, Wittgenstein, Malinowski, Jakobson and Sinclair, but are an area in which contemporary researchers ought to be dedicated and active. Table 2. Table showing the derivation of method and definitions in corpus stylistics Research
→
(Louw unless stated & chronological)
Taxonomy & Methodology
→
Glossary
Methodology preferred
→
entries
→
Close reading * Worlds and Symbolism 1987; *‘When + she + was’ (Sinclair, Phrasebite©) delexicalisation *Co-selection 1991; *All devices relexicalise Semantic Pros. *Delexicalisation is not 1993; metaphor First TaLC: *First move to science Terminology *Semantic pros. Binarity 1997; and science in exceptions, 1st falsification of *Semantic prosodies expanded(cpt) Halliday; *Collocation automates or 1999 (censored); falsfifies: paper for AACL 2008 Contextual (rejected: ‘conference full’) Prosodic * Approaches book revived by Theory (CPT): science Routledge, after its delay by Hoey. 2000; *Truth Studies (Truth and Reconciliation 2nd falsification Commission) TRC 2003; 1st mention philosophers TaLC6 2004 (published in 2007a); Worlds: Watson & Zyngier (Eds) Concordance instrumentation Sabotaged by large fonts 2006; Sinclair publishes Phrasebite© 2006; ACORN Keynote: Worlds and Wittgenstein
*worlds *co-select *delexical *relexical *metaphor *symbol *state of affairs *irony *insincerity *semantic prosody *device
(Continued)
Chapter 16. Consolidating empirical method in data-assisted stylistics
Table 2. Continued Research
→
Taxonomy & Methodology
→
Glossary
2007d; Relexicalisation Jakobson poetic function falsified Miller & Turci (Eds) 2007b Approaches to Corpus Stylistics (Routledge) delayed by Hoey.
This table is intended to show how the products of research do not automatically feed into methodology. The column on the left establishes the chronological order in which research is undertaken. The middle column ranks the findings from the first column in order of their importance to digital stylistics. The final column indicates how methodology needs to be anchored in a scientific form within a glossary of terms. Once this third stage has been reached, the glossary will need constant updating in order to incorporate new scientific findings as well as to exclude examples of poor science such as hybridization as well as methodological backsliding. If this procedure had been carried out after the death of J.R. Firth in December, 1960, the misleading reining back of collocation into syntax by Halliday (1966) might have been detected immediately and the damage it has caused to scholarship could have been avoided. One reason why semantic prosody is under constant attack (Hoey 2005, Whitsitt 2005, Hunston 2007) may be ascribed to the fact that its status as a scientific fact has not been settled in an appropriate corpus-attested glossary. Hunston (2007: 262) poignantly appears to miss the subtle relexicalization of point in the examples she puts forward as an apparent exception to the semantic prosody operating within to the point of. She offers to the point of gentleness, but without noticing the relexicalizing collocate fierce within the 9 word window, which stands ready to sharpen the point a little. The form is an exception because it is a device. 6. A scientific, corpus-attested glossary of literary terms If the products of research are not immediately converted into scientific practice, we can expect backsliding and hybridization to rein back our best digital research into analogue paths and methods. For this reason it is important that we offer guidance to stylisticians until the digital turn is fully established. Constant reinforcement of the manner in which literary devices operate as science is required as part of this process. The work of the hybridizers must be detected as quickly
Bill Louw
as possible and stamped out. Strenuous gate-keeping will be needed as part of this process of detection with a view to keeping falsifiable definitions and entries out of any proposed Glossary of Corpus-Attested Literary Terms. As the status of collocation as instrumentation becomes established, entries based upon intuition will gradually be banished, until all definitions contained in such a glossary are verifiable through the replication, using other or newer corpora, of the phenomena involved. This process will also safeguard all procedure against the homogenization of the corpus as an instrument (see Louw, forthcoming, for a comparison of the ‘sampling’ of the term natural justice in the British National Corpus and the Bank of English). This paper concludes by offering several draft entries for a potential glossary of corpus-attested literary terms. Successive editions of the glossary will tighten up the definitions in relation to the advancement of science alone, rather than the hopeful outpourings of theorists whose work is vitiated ab initio by being based upon the unverifiable tenets of psychologism. Hoey (2005) falls into this category. 6.1 Sample entries for a scientific glossary of literary terms The following sample entries are derived from some of the terms to be found in the third column of Table 2 above. They are not in alphabetical order for several reasons. Research findings cannot be classified by means of alphabetical order and the account of these advances are still in early draft. However, the final volume in which they appear will, of course, be organized in alphabetical order for ease of reference. Items in upper case in the entries cross-refer to other entries in the glossary.
Specimen Entries Worlds Literary worlds may be derived from Wittgenstein’s (1922: 7) assertion that: ‘The world is the totality of facts, not of things’. Facts are defined by Wittgenstein as ‘states of affairs’. Collocation and the technique of co-selection, between them automate Wittgenstein’s Picture Theory of Meaning and verify Sinclair & Maurenen’s (2006) Local Unit Grammar. This marks the point where textual and contextual ‘chunking’ meet. Where several states of affairs share a collocate, such as fire, the relexicalization of that collocate acts as a symbol whose involvement in several states of affairs, together sketch the literary world of the work in question. The world and literary criticism of it are further assembled on the
Chapter 16. Consolidating empirical method in data-assisted stylistics
basis of Wittgenstein’s (1922: 17) principle that states of affairs operate within logical space.
Irony Where a reversed semantic prosody is deliberate rather than inadvertent, we may assume that we are dealing with irony (Louw 1993). In a monolithically positive or negative semantic prosody, the exceptions characteristically occupy 3% of the total number of citations. Of these, two citations are usually ironic and the third (out of 100 citations) insincere. The latter are distinguished by the fact that they are uttered or written inadvertently.
Insincerity Where a reversed semantic prosody is inadvertent rather than deliberate, we may assume that we are dealing with insincerity. The binary principle obtaining between irony and insincerity was shown by Louw (2000) using the distinction between a + recipe + for and the + recipe + for. The author found that the insincere forms must occupy the form a + recipe + for + success as the reductio ad absurdum of a scientific proof. This was found to be the case without exception and within a sufficiently large sample of text to establish the scientific nature of semantic prosody (Louw 1993; 2000). In the case of insincere instances the text writes the attitude of its author (Louw 1993).
Metaphor Literary metaphor is usually the product of upward collocation (See the interview of John Sinclair by Wolfgang Teubert in the osti Report, Krishnamurthy, 2004: xxiii). The term metaphor is subject to confusion as a result of Lakoff and Johnson’s (1980; 2003) misleading description of what analogue stylistics might have termed ‘dead metaphor’. In reality, the metaphors described by these scholars are delexical forms rather than metaphors at all. They are the dead metaphors that form the bulk of ordinary language. Unless they are relexicalized, they do not qualify as devices and ought not to be referred to as metaphors and are useless to the pursuit of stylistics. The fact that Lakoff & Johnson refer to them as metaphors may well be a product of the fake relexicalization which takes place when examples are offered in the form of citation forms and out of context (See Frege 1884 & Philip 2003).
Bill Louw
Semantic prosody Two senses of semantic prosody are distinguished by Louw (1993; 2000). The first of these involves meaning through contagion by a collocate or set of collocates, through frequency, to impart a negative, positive or specialized meaning to a particular form. Specialized semantic prosodies are identified by Louw (1993) in the example fine, which is positive when applied to inanimate objects and ironic and negative in the case of animate persons such as friend. The substance of semantic prosodies is delexical, but in the case of irony and insincerity (created by means of breaching the semantic prosody) they are relexicalised, as are all devices. Sinclair (2004a) derives semantic prosodies structurally, by proceeding through linguistic levels in the normal direction associated with linguistic fieldwork. Louw (2000) by way of contrast derives them directly from Firthian and Malinowskian contexts of culture and contexts of situation, reversing the customary direction of linguistic field-work. Louw (2000) argues in his second and amplified definition of semantic prosody that they are a product of fractured contexts of situation: characterized by subtraction in the case of negative semantic prosody and by over-provision in the case of positive semantic prosody. The process involves seeing Firth’s taxonomy for context of situation and context of culture as being automated by collocation. Another name for this process is contextual prosodic theory. Semantic prosody has many detractors and hybridizers whose criticism is entirely intuitive and unsupported by corpus methods. Hoey (2005) and Whitsitt (2005) fall into this category.
7. Conclusion J.R. Firth once declared that scientific facts do not exist until they are stated. This is the common objective of all contributors to this volume. We ought not to allow instrumentation and the digital turn in stylistics to be defeated simply because we can see that it is at hand. If we are at last dealing with settled knowledge in stylistics, it needs, as Bertrand Russell advised, to be handed over to science. Philosophers and scientists seem to be more equal to this task than linguists and stylisticians. The allegiance of the latter to ‘schools of thought’ are likely for some time to come to be more of a hindrance than a help.
Chapter 16. Consolidating empirical method in data-assisted stylistics
References Abrams, M.H. 1957. A Glossary of Literary Terms. New York NY: Holt, Rinehart, Winston. Alvarez, A. 1964. The New Poetry. Harmondsworth: Penguin. Barthes, R. 1967. Elements of Semiology. London: Cape. Bazell, C.E. et al. (eds). 1966. In Memory of J.R. Firth. London: Longman. Belsey, C. 1980. Critical Practice. London: Methuen. Burge, T. 2005. Truth, Thought, Reason: Essays on Frege. Oxford: OUP. Cook, G. 2000. Language, Play, Language Learning. Oxford: OUP. Culler, J. 1975. Structuralist Poetics. London: Routledge and Kegan Paul. Enkvist, N.E. 1973. Linguistic Stylistics. The Hague: Mouton. Firth, J.R. 1957. Papers in Linguistics 1934–1951. Oxford: OUP. Fortunati, V. Forthcoming. Proceedings of the World Conference on Primitivism 2005. (to appear in Italian) Bologna. Frege, G. 1884/1974. The Foundations of Arithmetic: A Logico-Mathematical Enquiry into the Concept of Number, transl.by J.L. Austin. Oxford: Blackwell. Gardner, B. 1982. Up the Line to Death: The War Poets 1914–1918. London: Methuen. Halliday, M.A.K. 1966. Lexis as a linguistic level. In In Memory of J.R. Firth, C.E. Bazell et al. (eds), 148–162. London: Longman. Hoey, M.H. 2005. Lexical Priming: A New Theory of Words and Language. London: Routledge. Hunston, S. 2007. Semantic prosody revisited. International Journal of Corpus Linguistics 12(2): 249–268. Krishnamurthy, R. 2004. English Collocation Studies: The OSTI Report. London: Continuum. Lakoff, G. & Johnson, M. 1980/2003. Metaphors We Live By. Chicago IL: University of Chicago Press. Larkin, P. 1988. Collected Poems. Ed.by A. Thwaite. London: Faber. Louw, W.E. 1989. Sub-routines in the integration of language and literature. In Literature and the Learner: Methodological Approaches [British Council ELT Documents 130], 47–54. London: MEP. Louw, W.E. 1991. Classroom concordancing of delexical forms and the case for integrating language and literature. In Classroom Concordancing ELR Journal 4, T. Johns & P. King (eds), 151–178. Louw, W.E. 1993. Irony in the text or insincerity in the writer? The diagnostic potential of semantic prosodies. In Text and Technology: In Honour of John Sinclair, M. Baker, let al. (eds), 157–176. Amsterdam: John Benjamins. Louw, W.E. 1997. The role of corpora in critical literary appreciation. In Teaching and Language Corpora, A. Wichmann et al. (eds), 240–251. Harlow: Longman. Louw, W.E. 2000. Contextual prosodic theory: Bringing semantic prosodies to life. In Words in Context: In Honour of John Sinclair, C. Heffer & H. Sauntson (eds), 48–94. Birmingham: ELR. Louw, W.E. 2003. Dressing up waiver: A stochastic-collocational reading of the Truth and Reconciliation Commission (TRC). Harare: mimeo. Also available in the Occasional Papers dei Quaderni del CeSLIC at http://www.lingue.unibo.it/ceslic/e_occ_papers.htm. Louw, W.E. 2007a. Truth, literary worlds and devices as collocation. In Proceedings of the Sixth Conference on Teaching and Language Corpora, E. Hidalgo, L. Quereda & J. Santana (eds), 329–362. Amsterdam: Rodopi.
Bill Louw Louw, W.E. 2007b. Collocation as the determinant of verbal art. In Verbal Art Re-Visited, D. Miller & M. Turci (eds), 149–180. London: Equinox. Louw, W.E. 2007c. Corporibus Phlogiston: A Gentle Refutation of Michael Hoey’s Theory of Lexical Priming. Harare: Mimeo. Louw, W.E. 2007d. Are literary texts and their worlds ‘thrown together’ as collocation? Keynote presentation to ACORN Symposium, in Honour of John Sinclair, on 4th May 2007. www. aston.ac.uk/symposium.htm. Louw, W.E. 2007e. Literary worlds as collocation. In Literature and Stylistics for Language Learners: Theory and Practice, G. Watson & S. Zyngier (eds), 91–105. Basingstoke: Palgrave Macmillan. Louw, W.E. 2008 . Two chapters in Approaches to Corpus Stylistics, D. Hoover et al. London: Routledge. Macherey, P. 1966. Pour une Theorie de la Production Litteraire. Paris: Maspero. Malinowski, B. 1935. Coral Gardens and their Magic. London: Allen and Unwin. Miller, D. & Turci, M. (eds). 2007. Verbal Art Revisited. London: Equinox. Morrison, B. & Motion, A. (eds). 1982. Contemporary British Poetry. Harmondsworth: Penguin. Palmer, F.R. 1957/1968. Selected Papers by J.R. Firth 1952–1959. London: Longman. Philip, G.S. 2003. Collocation and Connotation: A Corpus-Based Investigation of Colour Words in English and Italian. PhD dissertation, University of Birmingham. Riffaterre, M. 1971. Essais de Stylistique Structurale. Paris: Flammarion. Sebeok, T.A. 1960. Style in Language. Cambridge MA: The MIT Press. Sinclair, J.M. et al. 1987/1993. Collins COBUILD English Language Dictionary. London: HarperCollins. Sinclair, J.M. 1987. Looking Up: An Account of the COBUILD Project in Lexical Computing. London: HarperCollins. Sinclair, J.M. 1991. Corpus, Concordance, Collocation. Oxford: OUP. Sinclair, J.M. 2004a. Trust the Text. London: Routledge. Sinclair, J.M. 2004b. Reading Concordances. London: Longman. Sinclair, J.M. 2006. Phrasebite. Pescia: TWC. Sinclair, J.M. & Mauranen, A. 2006. Linear Unit Grammar: Integrating Speech and Writing. Amsterdam: John Benjamins. Spitzer, L. 1948. Linguistics and Literary History: Essays in Stylistics. Princeton NJ: Princeton University Press. Watson, G. & Zyngier, S. (eds). 2006. Literature and Stylistics for Language Learners: Theory and Practice. Basingstoke: Palgrave Macmillan. Weber, J.J. 1996. The Stylistics Reader. London: Arnold. Whitsitt, S. 2005. A critique of the concept of semantic prosody. International Journal of Corpus Linguistics 10(3): 283–305. Wittgenstein, L. 1922. Tractatus Logico-Philosophicus. D.F. Pears & D.F. McGuiness (transl.), 1960. London: Routledge and Kegan Paul.
part iv
REDES Project: The New Generation
REDES Project: The new generation Introduction Instead of closing the book, the chapters in this section actually stand as the beginning, or ‘the new generation’ of international research. Willie van Peer’s work, as mentioned in the introduction to this volume, has influenced a number of people. Here, we would like to emphasize the example he has set for young researchers and scholars to be. In 2002, Willie van Peer co-founded the Research and Development in Empirical Studies Project (redes) with Sonia Zyngier and Frank Hakemulder at the Federal University of Rio de Janeiro. At that time, the project was triggered by a common desire to foster intercultural studies, building a bridge between the countries involved, namely, Germany, Brazil and the Netherlands. Moreover, the three co-founders believed students would gain from actively participating in research from the very early years of their university life, and, therefore, should be encouraged to develop their ideas and carry out their own studies. They assumed that, with proper supervision, these students would be better prepared for engaging in the academic world and would also be more aware of the stages involved in research (for a detailed account of the Project, see Viana et al. 2007). As stated in the Project’s philosophy, collaboratively written by its members in 2005, being a redes member “implies commitment, self-discipline, academic generosity, mutual respect and, in many cases, volunteer work” (Viana et al. 2007: 48). By now, several young researchers have had the experience of working in cooperation with students from different countries. Many papers have already been published (see Zyngier, Chesnokova & Viana 2007 for a collection of such articles), which stand as a proof that students can give relevant contributions to the academic community. No one expects them to undergo this experience relying solely upon themselves. On the contrary, being part of redes means holding hands and fighting battles together. This project aims at being more than an academic group. It is also a way of life, in which people are invited to act as a team, sharing research ideas, cultural backgrounds and personal experiences. redes members believe individuals contribute more to society when they are united in search for their goals. As a result, not only academic studies are carried out, but also bonds, both professional and personal, are made among people.
Directions in Empirical Literary Studies
In this section, the reader will find papers written basically by the new generation. Different from the previous sections, in this one there is no thematic thread. The papers stand as an illustration of how far junior researchers can go alone or in collaboration with senior members. The section opens with a chapter by Vander Viana, Natália Silveira and Sonia Zyngier, who represent the Brazilian branch of redes (see Viana 2007 for review). Their study presents an index of literariness based on three variables regarding the materiality of 46 literary works. With the help of a more objective method, the authors investigate how varied lexical items and their combinations are in each of the texts. They also check whether there is a correlation between the index of lexical variety they propose and the traditional subjective selection of a literary canon. From a more general perspective, the chapter shows how Corpus Linguistics can be used for the analysis of literature. Mariya Sergeyeva and Anna Chesnokova, redes members in Ukraine (see Sergeyeva & Rumbesht 2007 for the impact of the Project in this country), sign the second chapter. They focus on issues related to foreign language learning in their country. According to the main hypothesis of the study, the Russification or Ukrainization policies impact the way Ukrainians react to learning English. Ukrainian citizens’ attitudes are closely related to their previous school and academic experiences, whether in their native language or not. All in all, the paper points out to the importance of such aspects when learning a foreign language. Another representative of the Ukrainian redes Project Group, Vladimir Yepishev deals with the difficulties involved in translating proper names in The Lord of the Rings. As this is a multilingual trilogy, the names of the characters reflect their origin or have an impact on the plot. The author investigates how these names have been translated into Ukrainian and Russian. The text raises an important issue as to what should be taken into consideration when translating a work of art. In “North Korea’s ‘dangerous’ ways – reason for panic or a real threat?”, Jan Prasil, Maria Dudusova and Jan Auracher, members of the German branch of redes, investigate the influence of the media on the way citizens perceive North Korea. They focus on participants from different countries and the fear they might have after reading the official press release about the nuclear test held in North Korea. This text highlights the impact of the press on people’s assumptions of the world. The section closes with a chapter by Maciej Maryl from Poland. Although not officially a redes member since there is no branch of the Project in his country as yet, Maryl shares the spirit of redes, and his research has developed, in a way, because of his contacts with Willie van Peer. Maryl’s contribution focuses on noncanonical works and literary education. Although the study of literature in school settings is based on canonized literary works, the author reminds us most people rely on other kinds of reading in their daily routine. Therefore, Maryl reinforces the importance of non-canonical works in education.
REDES Project: The new generation
Inserting this section in the present volume shows that quality papers do not have to come together with academic prestige. More importantly, they pay a great homage to Willie van Peer. Such a section could not be left out from this Festschrift as Willie van Peer himself has been one of redes most enthusiastic members. By so doing, he has inspired many students, helping them through their projects and future careers in the university context. This has materialized in various ways, even when he is not aware of them as is the case with this volume. Milena Mendes Section Editor
References Sergeyeva, M. & Rumbesht, A. 2007. redes in Ukraine. igel Newsletter [online] 19. Available from http://www.arts.ualberta.ca/igel/Newsletter19.htm# Ukraine [Accessed on 01 April 2007]. Viana, V.P. 2007. redes-BRA: past, present and future. IGEL Newsletter [online] 19. Available from http://www.arts.ualberta.ca/igel/Newsletter19.htm# Redes-BRA [Accessed on 01 April 2007]. Viana, V.P., Fialho, O., Sopcák, P., Sergeyeva, M. & Rumbesht, A. 2007. (In)visible networks in action: Four perspectives. In Acting and Connecting: Cultural Approaches to Language and Literature, S. Zyngier, A. Chesnokova & V. Viana (eds), 23–49. Münster: lit Verlag. Zyngier, S., Chesnokova, A. & Viana, V.P. (eds). 2007. Acting and Connecting: Cultural Approaches to Language and Literature. Münster: lit Verlag.
chapter 17
Empirical evaluation Towards an automated index of lexical variety Vander Viana, Natalia Silveira & Sonia Zyngier This chapter proposes an objective approach to the formal analysis of literary prose in English in order to investigate the relation between lexical density and judgments of canonicity. Based on the concepts of literariness proposed by the Russian Formalists and lexical variety, a mathematical index is designed, relating three variables which take the materiality of text into consideration: (a) relative frequency of lexical bundles, (b) lexical bundle type/token ratio, and (c) word type/token ratio. The index is described and illustrated with 46 canonical and non-canonical literary works. Statistical analysis shows no significant relation between lexical richness and decisions of what has been classified as canonical, indicating that these judgments may be influenced by factors other than the text itself. Keywords: lexical variety, Corpus Linguistics, literary discourse, empirical study, canonicity, lexical bundles
1. Introduction Almost a century has gone by since the Russian Formalists proposed a systematized approach to literary studies, one which challenged the notion that the quality of literature could be decided by elements other than the text itself, to make a “stone stony” (Shklovsky 1917/1965: 12). Their seminal ideas of literariness, foregrounding, and deviation are the starting point of this paper. By assuming that the materiality of texts was the working ground for literary studies, the Formalists inaugurated the interface between literature and linguistics. Following this proposal, numerous investigations have resorted to methods of linguistic analysis to explore and explain literary texts, their properties, and their relations to other types of texts (e.g., Jakobson & Lévi-Strauss 1962/1987;
Vander Viana, Natalia Silveira & Sonia Zyngier
Fowler 1981; Carter & Simpson 1989; Carter & McRae 1996; Watson & Zyngier 2006, among others). What characterizes these publications is the departure from hermeneutics and other subjective techniques, and the promotion of replicable data analyses. The present study also breaks away from the hermeneutic tradition of literary studies by resorting to the tools provided by Corpus Linguistics, and thus offers a more objective look at textual features. Here lexical variety is taken as one important component of literariness and a selection of narrative prose in English is submitted to computerized analysis. The two research questions which guide the present study are (a) how varied the use of the lexicon in the works under scrutiny is and (b) whether there is any relation between lexical richness and canonicity.
2. Theoretical foundations 2.1 Russian Formalism One of the most important notions defined by Russian Formalism is that of defamiliarization, proposed by Shklovsky ([1917] 1965). The term implies viewers’ estrangement from an object and the effect provoked by prolonging their perceptions. This effect can be brought about by linguistic constructions which introduce novelty and challenge established norms. Still according to Shklovksy ([1917] 1965), the function of art is to lead individuals to a renovation of their understanding of the world through that sort of aesthetic experience. The process of defamiliarization, in which reflexive reactions are deautomated and called into question, allows viewers/readers to appreciate the beauty of an artistic object. By systematically resorting to it, a work of art comes to display density of forms rather than communicative fluidity. In the case of literature, defamiliarization is achieved through the manipulation of linguistic forms so as to produce foregrounding. The term “foregrounding” is used here in the sense of the stylistic devices that lead to the psychological effect described above (for other uses, see van Peer & Hakemulder 2006). It goes back to the distinction between foreground and background in visual arts: in the process of foregrounding, the defamiliarized object is called to the foreground, or “highlighted”, by means of two types of mechanism: deviation or parallelism. Deviation occurs when linguistic structures go against readers’ expectations, from what has been naturalized by the speakers of a language. They are deviations in relation to the grammatical rules and/or social conventions associated with the language, and the expectations such rules and conventions give rise to. In non-technical
Chapter 17. Empirical evaluation
contexts, deviation is usually referred to as poetic license. By applying deviational techniques, writers create a contrast between what is default and what is novel – or background and foreground. On the other hand, parallelism is a controlled form of repetition in which traces of a structure are kept constant while others vary. It is possible that all traces are repeated, thus accounting for simple repetition. However, it is the manipulation of variety and repetition that creates contrasts leading to stylistic effects and defamiliarization. It is worth noting, however, that the process of defamiliarization is countered by refamiliarization (see Miall & Kuiken 1994; Fialho 2007). Readers make use of refamiliarization strategies to interpret the difficult forms triggered by foregrounding, which for that reason tend to become worn out. The performance of foregrounding devices depends on their freshness. Therefore, the linguistic choices in text must be varied and constantly innovative in order to produce the effect of what the Formalists called defamiliarization. Taking these concepts into account, Russian Formalists, and later Czech Structuralists, put forward the notion that literary texts are distinguished by literariness, a property that results from the coherent and systematic use of the stylistic resources discussed above (Sklovsky 1917/1965; Mukarovsky 1979). Such texts are, to their view, marked by innovation and unexpectedness in the use of language. Nowadays, literariness is not considered a feature exclusive to literature, but perceivable in other genres, such as advertisements (Carter & Nash 1990; Cook 1992). It has come to be seen in a continuum rather than as dichotomic set (Carter & Nash 1983). By focusing on the text, Russian Formalists left us what is arguably their greatest contribution: the development of a new paradigm for literary studies, one which emphasizes textual verbal art, thus implying that there are no themes or social contexts which determine a priori what literature is. It is true that the Formalists concentrated on the language of poetry and avoided analyzing prose. In this study, however, their concepts will be applied specifically to the area of prose writing to examine the issue of lexical density. 2.2 Corpus Linguistics In order to carry out computer-aided analysis of prose texts, the present study applies the principles and practices of Corpus Linguistics. This knowledge area has been described as “the modern face of empirical linguistics” (Teubert 1996: vi) for
. For a more detailed account of novelty in art, see Martindale (this volume).
Vander Viana, Natalia Silveira & Sonia Zyngier
it analyzes language in use. It proposes that linguistic investigations should leave aside invented instances and explain language by means of naturally-occurring examples. This, however, does not mean intuitions should disappear completely, but they should rather only play a role in generating hypotheses and not in confirming them (for a historical overview of Corpus Linguistics, see Viana & Zyngier, forthcoming). Corpus Linguistics looks at language from a different perspective: it reveals patterns of language use, indicating that words do not occur at random. They attract or reject one another, forming patterns. Therefore, there is a need to check how words behave, especially if one assumes that form and meaning are inseparable (Stubbs 1993: 2). One of the most important points in Corpus Linguistics is that of probability in linguistic choices. Examining language in use may tell us how probable it is for a specific linguistic item to occur in a given register. For the sake of illustration, Louwerse, Benesh & Zhang (this volume) argue that the sequence “and in” is the most distinctive language feature of literary texts, thus possibly characterizing such type of discourse. In his seminal work Corpus, Concordance, Collocation, Sinclair (1991) proposes two principles to describe language use. For the author, writers and/or speakers are bound to them at all times. The open choice principle assumes users of a given language are (almost completely) free to choose the words they want to use. The only constraints they have to face are those dictated by grammatical rules. The idiom principle covers language use from another angle: speakers and/or writers use combinations of words they have already been in contact with before. In fact, Sinclair’s principles have had great impact. Louw’s (1993) concept of semantic prosody seems to be a clear example of that. In his words, it corresponds to “a consistent aura of meaning with which a form is imbued by its collocates” (Louw 1993: 157; also see Louw, this volume). Erman & Warren’s (2000) study shows that the idiom principle has an important role in language use as 55.38% of the choices made by both speakers and writers correspond to prefabricated structures. Wray’s (2002: 4) concept of formulaicity – “words and word strings which appear to be processed without recourse to their lowest level of composition” – also derives from Sinclair’s idiom principle. Similarly, Biber, Conrad & Cortes’s (2004: 372) statement that “much of our everyday language use is composed of prefabricated expressions” shows, in fact, Sinclair’s idiom principle at play. The authors forward the concept of lexical . Prefabricated structures are more common in spoken (58.6%) than in written registers (52.3%).
Chapter 17. Empirical evaluation
bundle – “the most frequent recurring lexical sequences in a register” (Biber, Conrad & Cortes 2004: 376), which had originally been proposed in Biber et al. (1999). In other words, a lexical bundle is a sequence of words that meet certain previously defined criteria. In the specific case of the study reported, the authors worked with four-word sequences that needed to occur in at least 5 distinct texts and a minimum of 40 times per million tokens. In classifying texts, Sinclair’s (1991) two principles may be interpreted as the two ends of a continuum. The least innovative texts, that is, the ones in which lexical repetition is commonly found, are the ones in which the authors rely the most on the idiom principle. At the other end, where one can find a heavier influence of the open choice principle, there are those texts in which the writers make more use of different words. The poles are hypothetical as it would be difficult to conceive a text in which there is no repetition or no new words. This proposal may be mapped onto the notion of literariness proposed by the Russian Formalists in the sense that higher literariness would derive from novel and unexpected language use, thus being a product of the open choice principle. On the other hand, a text characterized by repetition would stand as an example of lower literariness, being one in which the idiom principle has a strong influence. In the next section, these concepts will be put to practice.
3. Methodology For this study, a selection of 46 works of narrative prose in English was compiled, with first publication dates ranging from mid-18th century to late 20th century. It may seem at first that the broad time span, covering more than two centuries, would invalidate the present study. However, it should be stressed that although there is a comparison between the values obtained for each literary work under analysis, there is no cross-linguistic comparison among texts. Therefore, the time a literary work was produced does not skew the study. Out of the works analyzed, 27 are canonical (3 of which belong to children’s literature) while 19 are non-canonical.
. Repeated sequences of words have been studied by a number of scholars and labeled in different ways (see Wray 2002; Biber, Conrad & Cortes 2004). Louwerse, Benesh & Zhang (this volume) use the concept of ‘n-gram’. In this chapter, the notion of ‘lexical bundle’ is preferred since Biber, Conrad & Cortes’s (2004) procedures for identifying these sequences are also followed.
Vander Viana, Natalia Silveira & Sonia Zyngier
The literary works were transformed into text files, so as to allow later computer processing. First, some texts were manually digitalized from their printed editions, or, in the case of the public domain works, retrieved in digital form from Project Gutenberg. The resulting files were read, corrected and formatted in the second stage. Formatting standards were determined so as to avoid computer reading errors. Finally, the files were revised in order to ensure they were free from typos and/or other formatting problems. In relation to the units of analyses, both words and lexical bundles were employed. Words were identified as strings of letters and/or numbers with one space on each side (or a punctuation make to the left). Lexical bundles, as described by Biber, Conrad & Cortes (2004) (see Section 2.2), were identified as sequences of four words listed by the computer occurring at least 4 times per 100,000 tokens in a given text. All the 46 literary works were probed with WordSmith Tools (Scott 1999). This software offers three main tools: WordList, Concord and KeyWords. Here, WordList was used. This specific tool rendered data on the number of types (different words) and tokens (running words) of words and lexical bundles in each literary work, as well as the standardized type/token ratio of words (see Section 4 for an explanation of this ratio).
4. Index of lexical variety Lexical variety is investigated in this study by means of three variables relating to either lexical bundles or words. In the first case, two measures are considered: their relative frequency and type/token ratio. In the second, just the standardized type/token ratio is focused. The first variable looks into how frequent lexical bundles are in each literary work. What is considered, however, is the relative frequency of bundles, so as not to allow for text size bias. The value is calculated by dividing the raw frequency of lexical bundles by the number of tokens in the text. The result is then multiplied by 100,000 to ensure that all values are compared on the same basis, that is, number of lexical bundles per 100,000 tokens. Repetition of lexical bundles is less likely to occur than that of words, since it is defined by the use of the same four words in a fixed order. The more bundles are found in a text, the more repetitive it is and the higher the final value.
. Project Gutenberg is available at http://www.gutenberg.org/wiki/Main_Page. . In the original study (Biber, Conrad & Cortes 2004), the figures are different (40 times per 1,000,000 tokens), but the ratio is the same.
Chapter 17. Empirical evaluation
The second variable covers variety among the identified lexical bundles. What is at play in this measure is not how frequent bundles are (for it has been accounted for in the previous variable), but how different they are. This is the type/token ratio of lexical bundles. The value is arrived at by dividing the number of different bundles (types) by the relative frequency of bundles. The use of relative frequency instead of the raw frequency ensures that the final value is not biased by the size of the literary work. Besides the mere occurrence of bundles, which is investigated in the first variable, it should be taken into account how often they are repeated. It seems that there is more variety in the use of 20 bundles four times each than of 10 bundles eight times each. In sum, the more varied bundles are, the higher the value the text has in this variable and the richer it is in terms of its lexis. The third and last variable is similar to the second one, but it focuses on word level. It looks into the variety of isolated (rather than combined) words, which is expressed by the standardized type/token ratio of words. This measure is automatically calculated by WordSmith Tools (Scott 1999) every time a word list is run. The regular type/token ratio is obtained by dividing the number of types (different words) by the number of tokens (running words). The result is then multiplied by 100 so that the final value is expressed in percentage. However, this value is subject to the size of the text: the bigger it is, the higher the probability of it being lower. This is why the standardized type/token ratio is used. It functions in the very same way, except for the fact that the software calculates the ratio for each group of a certain number of tokens. By default, these groups are set as 1,000 tokens. At the end, the program calculates a total average, which corresponds to the final result. Here, the higher the final figure, the more varied words are. The index of lexical variety is a product of four steps. First, all values obtained in each of the variables are normalized. Normalization involves the following calculation: (x – minimal value in the variable) – (maximal value in the variable – minimal value in the variable) where x is the result obtained by a specific literary work. For instance, Alice in Wonderland and The Picture of Dorian Gray originally scored, respectively, 4,850.07 and 4.03 in the relative frequency of lexical bundles. With these results, the conclusion that could be arrived at was that Alice in Wonderland is more repetitive that The Picture of Dorian Gray; however, it would not be possible to state what the minimum and the maximum values of the continuum are, making it difficult to interpret the results qualitatively. After normalization is carried out, Alice in Wonderland scores 1.000 whereas The Picture of Dorian Gray totals 0.000. These two literary works represent the minimum and maximum . Since weights were assigned to the different criteria as will be explained later, the fact that their ranges (which originally had different orders of magnitude) were all normalized allowed for easy manipulation of the coefficients with which they were weighed.
Vander Viana, Natalia Silveira & Sonia Zyngier
values possible after the results are normalized. It should be stressed that 0.000 means Oscar Wilde’s literary work is the one that contains fewer lexical bundles, that is, it is less repetitive at word-sequence level in the selection of 46 works under analysis. Conversely, the one by Lewis Carroll resorts to lexical bundles the most. Second, weights are attributed to each variable. It seemed senseless to treat all variables in the same way as two are based on lexical bundles while one is based on isolated words. After all, the probability of repeating bundles is lower than that of using the same words. A number of empirical tests were carried out, assigning different weights to the variables, so as to calibrate the index. In choosing the best design possible, it was assumed that James Joyce’s Ulysses should be among the first results (if not the first one) since it is known that this literary work is lexically rich and innovative. As a result, a double weight was attributed to the first two variables when compared to the third variable. In other words, the relative frequency of lexical bundles and their type/token ratio were assigned 2 whereas the standardized type/token ratio was assigned 1. However, as the relative frequency of lexical bundles differed from the other variables – a higher result would indicate more repetition – it was assigned a negative weight: –2. Third, the normalized results obtained in each literary work for each variable are multiplied by their respective weights and added up. The final result would correspond to the index of lexical variety. The most varied literary work would score 0.000, 1.000 and 1.000 in, respectively, relative frequency of lexical bundles, type/token ratio of lexical bundles and standardized type/token ratio of words, totaling 3.000. On the other end of the continuum, there would be a literary work scoring –2.000, 0.000 and 0.000, corresponding to –2.000. The fourth and last step in the calculus design is a consequence of the result obtained in the previous one. Were the index kept to its original form, it would range from –2.000 to 3.000. Thus, it could raise unwanted interpretation for negative results, i.e., that a literary work would be so repetitive that it would owe something to “default” language use. This is clearly not the case, especially because the index is relative and its values may change every time new works are included in it (although the relation among them is not affected). For this reason, it was decided to make the results positive by adding 2 to all of them. This way, the final index ranges from 0.000 to 5.000.
5. Analysis Each of the selected 46 literary works was subjected to the procedure described in the previous section so as to arrive at the index of lexical variety. Table 1, which shows the results of the analysis, contains five columns. The first one indicates the
Chapter 17. Empirical evaluation
titles. The use of italics indicates a popular literary work whereas bold type signals a canonical one. The three asterisks after certain canonical literary works (in bold type) indicate that they are aimed at children. The second column [rf(B)] corresponds to the relative frequency of lexical bundles. The following two indicate type/token ratios: ttr(B) indicates the ratio for lexical bundles whereas sttr(W) stands for the standardized ratio for words. The last column (ILV) contains the final result for the index of lexical variety. The last row shows the weights which were assigned to the three variables. The 46 literary works are arranged according to their ILV in descending order. Table 1. Literary works and their indexes of lexical variety
Work
rf(B)
ttr(B)
sttr(W)
ILV
Ulysses The Picture of Dorian Gray Jane Eyre Wuthering Heights Moby Dick Noble House The Beautiful and Damned Tess of the d’Urbervilles Nicholas Nickleby You Only Live Twice The Blind Assassin Bleak House The Mirror of the Sea The Voyage Out Live and Let Die The Handmaid’s Tale Pride and Prejudice The Da Vinci Code The Dark Half Great Expectations Thunderball The Secret of Chimneys David Copperfield On Her Majesty’s Secret Service Harry Potter and the Chamber of Secrets Hard Times Harry Potter and the Goblet of Fire The Garden Party and Other Stories The War of the Worlds Sons and Lovers
0.005 0.000 0.013 0.018 0.018 0.024 0.032 0.022 0.032 0.142 0.033 0.030 0.137 0.084 0.195 0.111 0.097 0.134 0.091 0.055 0.099 0.241 0.044 0.110 0.142 0.051 0.116 0.139 0.212 0.092
0.697 1.000 0.742 0.629 0.460 0.508 0.367 0.395 0.438 0.490 0.409 0.500 0.404 0.404 0.493 0.454 0.481 0.341 0.410 0.456 0.326 0.532 0.404 0.321 0.300 0.335 0.325 0.344 0.337 0.347
1.000 0.343 0.598 0.808 0.716 0.608 0.768 0.605 0.506 0.619 0.561 0.310 0.680 0.535 0.558 0.467 0.325 0.663 0.418 0.241 0.580 0.444 0.285 0.573 0.612 0.354 0.465 0.403 0.546 0.285
4.383 4.343 4.057 4.030 3.600 3.575 3.437 3.351 3.318 3.314 3.313 3.251 3.213 3.175 3.154 3.152 3.092 3.077 3.057 3.043 3.034 3.026 3.006 2.996 2.927 2.923 2.883 2.812 2.796 2.795 (Continued)
Vander Viana, Natalia Silveira & Sonia Zyngier
Table 1. Continued
Work
rf(B)
ttr(B)
sttr(W)
ILV
The Adventures of Sherlock Holmes The Spy Who Loved Me Harry Potter and the Sorcerer’s Stone Our Mutual Friend The Portrait of a Lady Diamonds are Forever Peril at End House The Invisible Man Casino Royale The Wedding The Adventures of Huckleberry Finn The Notebook The Jungle Book*** Murder On The Orient Express A Christmas Carol*** Alice in Wonderland***
0.109 0.589 0.136 0.041 0.077 0.346 0.169 0.596 0.166 0.185 0.168 0.811 0.265 0.403 0.381 1.000 -2
0.320 0.766 0.274 0.270 0.271 0.432 0.328 0.674 0.204 0.291 0.386 0.736 0.159 0.268 0.143 0.000 2
0.373 0.424 0.497 0.307 0.316 0.523 0.259 0.406 0.480 0.328 0.000 0.226 0.211 0.244 0.393 0.000 1
2.795 2.778 2.773 2.764 2.703 2.695 2.577 2.562 2.557 2.539 2.436 2.074 1.998 1.974 1.917 0.000
The most varied literary work in the list, as expected, is Ulysses, scoring 4.383. At the end of Table 1, there is Alice in Wonderland, which scores 0.000 in the index, having obtained the lowest result possible in the three variables. Even though Alice in Wonderland is considered a canonical literary work, Lewis Carroll writes it for children. This may explain why there are numerous repetitions in the book and why it has achieved the minimum result possible. Table 1 may be divided into three segments. The upper one, containing 15 literary works, seems to mostly group canonical ones. There are only three works which are popular in this initial segment: Noble House, You Only Live Twice and Live and Let Die, two of which are of the same author – Ian Fleming – who signs the James Bond series. The last segment of 15 literary works spans from The Spy Who Loved Me to Alice in Wonderland. It mainly encompasses non-canonical literary works (8 instances) and children literature (3 instances). The appearance of children’s literature at the bottom of the list was already expected. Such texts are generally more repetitive than works written for adults, and it is reasonable to assume that their
. It should be explained why there is not a literary work which obtains 5.000 in the index. This would correspond to that literary work which obtains the maximum results in each of the variables. So far this does not hold true for the analyzed texts. The Picture of Dorian Gray yields the best results in terms of bundles (the first two values in Table 1), respectively 0.000 and 1.000. However, it is Ulysses that scores highest in the third variable, obtaining 1.000.
Chapter 17. Empirical evaluation
authors resort to repetition as a means of appealing to the readers they address. Only four canonical works may be found here: Our Mutual Friend, The Portrait of a Lady, The Invisible Man and The Adventures of Huckleberry Finn. It is interesting to point out that this last work – one in which an attempt was made to particularize the use of the vernacular in the Southern United States – repetition seems to have been a tool in the characterization of that use of language. In the middle part, there are 16 literary works, which are evenly distributed. One half consists of canonical works while the other encompasses popular ones. In order to check the possible significant relation between canonicity and the index of lexical variety, a correlation test was run for the distribution. A dichotomous variable ‘canonicity’ was created, and a value {1,0} was assigned to each work. That variable was correlated to ‘lexical variety’, the interval variable corresponding to the value each work obtained in the index. All 46 cases were considered. A Pearson correlation test yielded r = 0,132 for p = 0,190 with a one-tailed test, which means that no significant correlation was found. This finding indicates that there is not a strong relation between lexical richness, an important formal aspect of every text – and one interpreted here, within the theoretical framework of Russian Formalism, as an indicator of literariness – and judgments of canonicity. 6. Final words Grounded on the notion of literariness and helped by the principles and tools of Corpus Linguistics, this study aimed at advancing in the investigation of lexical variety instead of relying only on the common standardized type/token ratio of words. This way an index of lexical variety was proposed considering not only words but also lexical bundles. The index showed how rich in terms of language use each literary work is. Canonical works seemed to be higher on the list while popular texts clustered towards the end. In the case of canonical works for children, the results showed they drifted to the end of the list. However, when the relation between lexical richness as modeled here and canonicity was tested statistically, results indicated that the correlation was not strong. This may lead to the conclusion that judgments of canonicity seem to be further distanced from textual aspects than usually assumed. In other words, the inclusion or exclusion of literary works from the canon of a specific culture may challenge the notion that textual features are strong indicators of canonicity. As far as the index is concerned, a number of aspects still need to be mapped. For instance, it does not capture different or new meanings of the same word, nor does it take metaphors into account. New variables need to be made operational. Other kinds of analyses can also enrich the final result. The rapid development in the field of Corpus Linguistics will possibly offer new ways to improve this index.
Vander Viana, Natalia Silveira & Sonia Zyngier
References Biber, D., Conrad, S. & Cortes, V. 2004. If you look at…: Lexical bundles in university teaching and textbooks. Applied Linguistics 25(3): 371–405. Biber, D., Johansson, S., Leech, G., Conrad, S. & Finegan, E. 1999. Longman Grammar of Spoken and Written English. London: Longman. Carter, R.A. & McRae, J. 1996. Language, Literature and the Learner. London: Longman. Carter, R.A. & Nash, W. 1983. Language and literariness. Prose Studies 6(2): 121–141. Carter, R.A. & Nash, W. 1990. Seeing through Language: A Guide for Styles of English Writing. Oxford: Basil Blackwell. Carter, R.A. & Simpson, P. (eds). 1989. Language, Discourse and Literature: An Introductory Reader in Discourse Stylistics. London: Unwin Hyman. Cook, G. 1992. The Discourse of Advertising. London: Routledge. Erman, B. & Warren, B. 2000. The idiom principle and the open choice principle. Text 20: 29–62. Fialho, O.C. 2007. Foregrounding and refamiliarization: understanding readers’ response to literary texts. Language and Literature 16(2): 105–124. Fowler, R. 1981. Literature as Social Discourse: the Practice of Linguistic Criticism. London: Batsford Academic. Jakobson, R. & Lévi-Strauss, C. [originally published in 1962] 1987. Charles Baudelaire’s “Les Chats”. In Language in Literature, K. Pomorska & S. Rudy (eds), 180–197. Cambridge MA: Belknap. Louw, B. 1993. Irony in the text or insincerity in the writer? – The diagnostic potential of semantic prosodies. In Text and Technology: In Honour of John Sinclair, M. Baker, G. Francis & E. Tognini-Bonelli (eds), 157–176. Amsterdam: John Benjamins. Miall, D.S. & Kuiken, D. 1994. Foregrounding, defamiliarization, and affect: Response to literary stories. Poetics 22: 389–407. Mukarovsky, J. 1979. Aesthetic Function, Norm and Value as Social Facts. Oxford: Oxon. Scott, M. 1999. WordSmith Tools 3.0. Oxford: OUP. Shklovsky, V. 1917/1965. Art as technique. In Russian Formalist Criticism: Four Essays, L.T. Lemon & M.J. Reis (eds & transl), 3–24. Lincoln NE: University of Nebraska Press. Sinclair, J. 1991. Corpus, Concordance, Collocation. Oxford: OUP. Stubbs, M. 1993. British traditions in text analysis: From Firth to Sinclair. In Text and Technology: In Honour of John Sinclair, M. Baker, G. Francis & E. Tognini-Bonelli (eds), 1–33. Amsterdam: John Benjamins. Teubert, W. 1996. Editorial. International Journal of Corpus Linguistics 1(1): iii–x. Van Peer, W. & Hakemulder, J. 2006. Foregrounding. In Encyclopaedia of Language and Linguistics, K. Brown (ed.), 546–551. Oxford: Elsevier. Viana, V. & Zyngier, S. Forthcoming. EFL through the digital glass of Corpus Linguistics. In Handbook of Research on E-Learning Methodologies for Language Acquisition, P.L. Torres & R. de C.V. Marriott (eds). Hershey PA: IGI Global. Watson, G. & Zyngier, S. (eds). 2006. Literature and Stylistics for Language Learners: Theory and Practice. Hampshire: Palgrave Macmillan. Wray, A. 2002. Formulaic Language and the Lexicon. Cambridge: CUP.
chapter 18
Language allergy Myth or reality? Mariya Sergeyeva & Anna Chesnokova The chapter examines the impact of Russification and Ukrainization policies on the attitude of Ukrainians towards foreign (English) language learning. The hypothesis is that people who had a negative experience with instruction in their non-native language are likely to experience the same reaction when learning another foreign language. To test this hypothesis, 96 participants filled in questionnaires to evaluate their experiences with studying in non-native languages and with English language learning. Results indicate that there is a strong correlation between experiences with instruction in non-native language and with learning foreign languages. Keywords: language learning, motivation, Russification, Ukrainization, language allergy, empirical study
1. Introduction With Ukraine in the process of building solid relations with, or even becoming a member of Western organizations such as the Council of Europe, the European Union, osce, wto, and preparing to become the host of world sports events (e.g., “Eurocup” 2012), awareness of the need for multicultural policies in this postSoviet country is rising fast. Among others, the issue of foreign language learning (English in particular, as lingua franca for international communication) is gradually receiving greater attention as an essential factor for becoming a member of the European and world community.
. For explanation of difference between “Ukraine” and “the Ukraine”, please refer to http:// www.infoukes.com/faq/the_ukraine/ and http://www.wsu.edu/~brians/errors/ukraine.html
Mariya Sergeyeva & Anna Chesnokova
According to the Eurobarometer survey in 2001, while “47% of eu citizens spoke English well enough to hold a casual conversation” (European Commission 2001), no such figures exist in Ukraine – there are no statistical data on this issue. However, very rough figures gathered by means of a street-poll show that approximately only 1 Ukrainian out of 5 speaks English on a level sufficient for general communication. The street-poll was carried out in September 2007 by a Portuguese ex-patriot on one of the central streets of Kyiv. Twenty-eight randomly chosen people were asked in English to give directions to different places. Only 4 people showed an excellent knowledge of English, and 2 people demonstrated an ability to communicate in English sufficiently well. It is reasonable to assume that if the street-poll had been carried out in any other Ukrainian city, the figures would have definitely been much lower. Such wide discrepancy between knowledge of English in Western European countries and in Ukraine triggered the question of what factors are accountable for the lack of English language knowledge among the Ukrainian population. 2. Why don’t Ukrainians speak English? According to the data collected in the Eurobarometer survey, there are 3 factors mentioned by the European respondents as discouragement from language learning: “lack of time (34%), motivation (30%), and expense of language classes (22%)” (European Commission 2006: 5). Analyzing the factors accountable for the lack of knowledge of English among the Ukrainian population, we distinguished the following factors: 1. 2. 3. 4. 5.
educational motivational geopolitical environmental emotional.
2.1 Educational factor In Ukraine, English language learning starts in primary school and continues through higher education. The Ukrainian system of education is rather new and does not comprise members of the generation who studied more than 16 years ago. Thus, the Soviet system of foreign languages education should also be taken into consideration. At that time, learning English as a foreign language usually started in the secondary school and continued at higher educational institutions. Without evaluating the Soviet system of English language instruction, one may expect that more than 10 years of even poor language learning may result
Chapter 18. Language allergy
in a level of English language proficiency that is sufficient for communication. Nevertheless, the figures presented earlier in this chapter show the opposite. 2.2 Motivational factor These figures may be explained by the lack of incentives, material in particular. The study of the motivation for learning languages in the European Union shows that the reasons for foreign language learning are becoming more and more tied to practical benefits, such as using the skills at work (32%), working abroad (27%), and getting a better job inside the country (23%) (Special Eurobarometer 2006: 5). The analysis of 50 randomly chosen employment ads (taken from www.alljob.com. ua), all offering a salary higher than $1,000, however, shows that 32 of them (about 65%) require a “good knowledge of English”. The examination of 50 randomly chosen resumes from the same Internet source demonstrates that only 14 (fewer than 30%) job-seekers meet this requirement, indicating “fluent English” in their CVs. If we compare the aforementioned salary with 1,475 uah (about $292), the official average wage in Ukraine in October 2007 (Average wages and salaries by region 2007), it triggers the question: is this not a sufficient motivation for learning English? This question is especially relevant taking into account the data compiled by the Ukrainian Institute of Social Studies and “Social Monitoring” Center. According to these data, 92% of Ukrainian youth give priority to “high level of payment” in their system of values (Levkovska 2006). 2.3 Economic factor Having mentioned that money may be a potential motivation for language learning, it may also be considered an obstacle one must accept to learn a language. To test this assumption, we gathered information about English group tuition fees in different language courses in Kyiv, with the result that the average monthly cost per person varies from 450 uah to 650 uah ($90–$130), the cost per hour being 18 uah-27 uah ($3.5–$5). It is remarkable that a 1.5–hour lesson of English in Kyiv costs less than a cinema ticket, priced 40 uah ($8), thus giving us the grounds to claim that the economic factor is not a sufficient reason for the lack of English language knowledge among Ukrainians. 2.4 Geopolitical factor Despite the fact that geographically Ukraine may be considered the center of Europe, politically this is by far not the case. In this respect, it is essential to mention that until 16 years ago (the collapse of the Soviet Union in 1991), the borders of Ukraine, as of all the other Soviet Republics, were closed to Western influence, which resulted in the absence of an English language environment
Mariya Sergeyeva & Anna Chesnokova
that would make the learning process faster and easier. Nonetheless, although the cultural influence of Western civilization (the United States in particular) on Ukraine has sky-rocketed since 1991, the level of English language knowledge has not developed at such a rate. 2.5 Emotional factor Much of the research of the last decades has concentrated on the role of emotion in learning (Christianson 1992), though researchers have different, often quite contradictory, opinions on this issue, with some of them claiming that “the memory advantage for highly affective material was the same for negative as for positive reactions to the material” (Bower 1992: 15), and others stating that it is only positive affect that has a powerful and facilitating effect on memory and learning (Isen 1999). Taking the aforementioned opinions into account, investigation of what emotions, if any, affect learning English in Ukraine seems reasonable. Can emotions enhance or hinder this process? Is it necessarily the case that positive emotions account for success in acquiring skills in English, and negative ones are to blame in failing to do so? Is there such a notion as “language allergy”? Is it “language allergy” that prevents Ukrainians from learning English? These questions cannot be answered without some awareness of the linguistic situation in Ukraine and its language policies during the 20th century.
3. Linguistic situation in Ukraine: Historical background As the interest towards bilingualism and its impact on education in post-Soviet countries is growing (Batelaan 2002), we decided to take a closer look on the attitude to it in Ukraine. Since Ukraine became an independent state in 1991, Ukrainian has been the only official state language, though issues in bilingualism have always been on the political agenda (Kolesnikov 2003). One of the key slogans of a leading political party was the idea of introducing Russian as a second state language, which raised numerous discussions both during the last presidential and parliamentary elections (Preobrazhenskaya 2006). There were two controversial views on this issue. The opinion shared by the majority of the Western Ukraine population was that bilingualism would lead to another Russification of Ukraine, thus restricting the rights of the Ukrainian-speaking population in their own country (Slezko 2000; Shaizhin 2006; Podobed 2007). Another view ascribed mostly to the population of the Eastern Ukraine stated that bilingualism is essential for preserving the rights of those who consider Russian as their native language (Kotsina 2007).
Chapter 18. Language allergy
Before taking a closer look at the linguistic processes in Ukraine, the concepts of Ukrainization and Russification need to be clarified. While Russification is considered as a “forcible introduction of Russian language and culture” (Busel 2006: 1279), Ukrainization is believed to be the “introduction of Ukrainian language, culture, and traditions” (idem: 1502). Having stated above that bilingualism is a highly political issue, we shall not attribute either positive or negative assessment to either of the processes. Thus, as an inventory for the research, we will refer to Russification and Ukrainization as historical periods when secondary and higher education was mainly carried out in Russian or Ukrainian respectively. We shall present the data testifying to fast interchangeability of Ukrainization and Russification processes, which may have resulted in the rejection of the governmental language policy by the population of Ukraine. In 1958, only 21% of children studied in Ukrainian schools compared to 1926, when over 97% of high school students were obtaining their education in Ukrainian (Dzyuba 2005: 175). It is worthy of note that, before 1920s, when no statistical data were available, there were hardly any Ukrainian schools at all after the Ems Ukaz of 1876, which banned the Ukrainian language. In 1990–1991 the number of students using Ukrainian in higher educational institutions of Ukraine constituted only 7% (Ivanishin and RadevychVynnizkiy 1992: 128). This situation changed dramatically, however, after 1991 when the independence of Ukraine was proclaimed and Ukrainian was recognized as the only official state language. Consequently, during the first years of independence, the distribution of students by Ukrainian language of instruction was as follows: secondary educational institutions: 1991–92 –45%, 1995–96 –58%, 2004–05 –77%. In higher educational institutions of I and II level of accreditation there were 7% of students studying in Ukrainian in 1990–91. In 1995–96 these numbers soared to 55% and in 2005–06, to 85% (Statistical Yearbook of Ukraine 2004: 477; Statistical Yearbook of Ukraine 2007: 484). These figures illustrate the fact that over the first decade of independence the government transformed the system of secondary and higher education from partly to overwhelmingly Ukrainian. These essential, though abrupt, changes led to the charges of Ukrainization, put forward mostly by the Russian-speaking population.
4. Hypothesis and methodology With such a long history of sudden changes in language policy taking place every 10–20 years, it is only natural to assume that there should be some kind of opposition from the population towards this policy. What remains to be proved, however, is whether this opposition, or “language allergy”, hinders learning other foreign languages, English in particular, through internal resistance towards learning.
Mariya Sergeyeva & Anna Chesnokova
The current study investigates the possible impact of Russification and Ukrainization policies on the attitude of Ukrainians towards foreign, in particular English, language learning. The paper attempts to explore the degree of correlation, if any, between the experiences of being instructed in non-native language, Russian or Ukrainian, in secondary or higher educational institutions and English language learning. The hypothesis is that people who had negative experience with instruction in their non-native language are likely to experience the same reaction when learning another foreign language. To test this hypothesis, 96 participants responded to a questionnaire. They came from 3 regions: 31 from Kyiv, the bi-lingual capital, 36 from the Western, mostly Ukrainian-speaking region, and 29 from the Eastern, Russian-speaking region. Their ages ranged from 17 to 52 with a mean of 27.96. The choice of participants was random. The respondents filled in the questionnaires during their business activities in offices and during leisure activities in public places in a fully anonymous way. In the first part of the questionnaire, the respondents indicated their age, gender, occupation, native language, language of instruction at their secondary and higher education institutions. In addition, they stated whether they had ever learnt the English language and evaluated their knowledge of English. Whether these evaluations were subjective or objective did not affect the course of investigation, as it is the attitude towards English language learning and self-evaluation that were of prime importance for the research. In the second part of the questionnaire, the participants ranked their attitude towards being instructed in their non-native language, Russian or Ukrainian, with the help of the adjectives on a semantic differential scale. The terms on this scale were selected during the preparatory stage by means of free associations with “instruction in the non-native language” of 38 randomly chosen respondents. Six adjectives and their opposites were chosen as the most frequently mentioned: “easy/difficult”, “interesting/dull”, “exciting/boring”, “useful/useless”, “pleasant/unpleasant”, “optional/forced”. The respondents also evaluated their experiences with learning English or their opinion about learning English if they had not studied it. To do so, they used the adjectives also selected at the preparatory stage by means of free associations with “English language learning” of the same 38 randomly chosen participants. Six adjectives and their opposites were selected on the basis of both their frequency and their compliance with the previous set of adjectives: “easy/difficult”, “interesting/dull”, “exciting/boring”, “useful/useless”, “possible/impossible to learn”. It should be pointed out that though the pairs “interesting/dull” and “exciting /boring” are quite close in meaning, both of them were selected to serve the filter
Chapter 18. Language allergy
against the automatically-filled questionnaires. Thus, if the evaluations with the help of these pairs of adjectives contradicted each other, the questionnaire would be considered invalid. Finally, participants were divided into 3 groups according to their experience with non-native language instruction in secondary or higher educational institutions. The first group, named “native” and functioning as control group, included the respondents whose language instruction had been in their native language, and who therefore had not undergone any influence of non-native language instruction. The second group consisted of respondents who had had a positive experience with instruction in a non-native language. This group was named “non-native positive”. The third group, named “non-native negative” comprised respondents who had had negative experiences with instruction in a non-native language. 5. Analysis and discussion For the quantitative analyses of the data, spss for Windows version 11.0 was used. anova was used for comparing the responses of the 3 groups of participants. Before moving on to the analysis of the results, the spread of respondents should be pointed out. As an example we will take the “pleasant/unpleasant” category:
non-native unpleasant
native
non-native pleasant
Graph 1. Groups division according to “pleasant/unpleasant” experiences with non-native language instruction.
Mariya Sergeyeva & Anna Chesnokova
As we can see from Graph 1, the group of respondents who had had unpleasant experiences with non-native language of instruction constitutes 33% of the respondents, which is enough to wield considerable influence on the overall picture of the attitude towards English language learning. If we consider the distribution of these 3 groups across the regions of Ukraine, we obtain quite remarkable results. Graph 2 shows that in all the regions of Ukraine the percentage of people who represent “native” or “control” group is virtually the same. That brings us to the conclusion that people who were not exposed to non-native language of instruction in their secondary or higher institutions are distributed equally across Ukraine. Thus, people who were exposed to a non-native language of instruction are also distributed equally across the country. However, the distribution of people who had positive and negative experiences with non-native language instruction, “non-native pleasant” and “non-native unpleasant” groups, respectively, differs considerably. In the first region, Kyiv, there are twice as many respondents who evaluate their experiences with non-native language positively than in the second, and especially the third,
60
50
Percent
40
non-nat lang pleasant
30
20
native non-native pleasant
10
non-native unpleasant 0
Kyiv
East West place of birth
Graph 2. Groups distribution across Ukraine according to “pleasant/unpleasant” experiences with non-native language instruction.
Chapter 18. Language allergy
Eastern and Western regions respectively, where there are considerably more respondents who constitute the “non-native unpleasant” group. Though not highly significant (p = 0.056), these results trigger a few practical questions. Who are these people in the regions who consider their experience with non-native language of instruction negative? What makes Kyiv region’s system of education so different from other Western and Eastern regions that results in considerably higher percentage of respondents having positive experiences with nonnative language instruction? The answer to these questions may serve as one of the instruments in facilitating English language learning in Ukraine, due to the fact that highly significant correlations between the experiences of being instructed in non-native language and of English language learning were found. As we can see from Graph 3,2 the evaluation of English language learning by “native” or control group differs substantially from that of the “non-native unpleasant” group that has much lower percentage of respondents who give positive evaluations of English language learning and higher percentage of those who give 50
40
Percent
30 non-nat lang pleasant 20 native 10
0
non-native pleasant non-native unpleasant
Graph 3. Distribution of respondents in “pleasant/unpleasant” category of non-native language of instruction and English language learning.
2. The horizontal line at the bottom of Graphs 3 to 6 moves from positive to negative, indicating a scale from “very+, a bit+, neither/nor, a bit–, very –”.
Mariya Sergeyeva & Anna Chesnokova
negative evaluations than the control group (p = .000). There is a clear trend of the “non-native unpleasant” group to give indifferent evaluation (“neither pleasant nor unpleasant”) to English language learning. The same results, also highly significant (p = 0.002), can be traced while examining the correlation between the experiences in a non-native language of instruction in the category “unforced/forced”, and the “pleasant/unpleasant” category of English language learning (Graph 4).
50
40
Percent
30 non-nat lang forced 20 native 10
non-native unforced non-native forced
0 Graph 4. Distribution of respondents in “unforced/forced” category of non-native language of instruction and “pleasant/unpleasant” category of English language learning.
It is clearly seen that respondents who consider their non-native language of instruction as “forced” tend to evaluate English language learning as “pleasant” less frequently and as “unpleasant” more frequently than the control group, again giving preference to indifferent evaluation. Yet, quite unexpected results were produced by examining the correlation between the experiences in a non-native language of instruction in the category “unforced/forced” and “interesting/dull” (p = 0.003) (see Graph 5).
Chapter 18. Language allergy 70 60
Percent
50 40 non-nat lang forced 30 20 10 0
native non-native unforced non-native forced
Graph 5. Distribution of respondents in “unforced/forced” category of non-native language of instruction and “interesting/dull” category of English language learning.
The same result can also be seen between the experiences with a non-native language of instruction in the category “pleasant/unpleasant” and “interesting/dull” (p = 000) (see Graph 6). 80
Percent
60
40
non-nat lang pleasant native
20
non-native pleasant non-native unpleasant
0 Graph 6. Distribution of respondents in “pleasant/unpleasant” category of non-native language of instruction and “interesting/dull” category of English language learning.
Mariya Sergeyeva & Anna Chesnokova
The trends between the control group and the “non-native forced” group (Graph 5), which is the group of respondents who consider their non-native language of instruction as forced upon them, as well as the control group and the “non-native unpleasant” group (Graph 6), confirm the tendencies shown in Graphs 3 and 4. It is worthy of note, though, that both the “non-native forced” group and the “non-native unpleasant” group almost do not give indifferent evaluation of English language learning in “interesting/dull” category as opposed to Graphs 3 and 4, where indifferent evaluation is far more frequent. What is even more remarkable, however, is that the percentage of respondents from the “nonnative unforced” and “non-native pleasant” groups that evaluate their experience in English language learning as “interesting” is higher than that of the control group.
6. Conclusions Having considered all the categories, we can conclude that the hypothesis was confirmed in the following cases: ■ ■
■
■
respondents who had unpleasant experiences with non-native language of instruction tend to have unpleasant experiences with English language learning; participants who consider that non-native language of instruction was forced upon them have a greater tendency to evaluate learning English as unpleasant than other groups (control and “non-native pleasant”); those who believe that they were forced into studying in their non-native language evaluated English language learning as dull more frequently than other two groups; the same trend is preserved with respondents who had unpleasant experience with being instructed in non-native language – they also state that learning English is dull as opposed to control and “non-native pleasant” groups.
Taking into consideration the results obtained from examining the aforementioned categories, we may infer that Ukrainians tend to evaluate English language learning as something interesting, but not very pleasant. Thus, as regards the question of whether Ukrainians have “language allergy”, no definite answer can be given, despite the high level of correlation between some experiences of being instructed in non-native language in secondary or higher educational institutions and English language learning. The results of the present study may serve as a basis for further research on the methods used for the teaching of English in Ukraine.
Chapter 18. Language allergy
References Average wages and salaries by region (monthly information). 2007. Retrieved November 16, 2007, from http://www.ukrstat.gov.ua/ Batelaan, P. 2002. Bilingual education: The case of Latvia from a comparative perspective. Intercultural Education 13(4): 355–374. Bower, G. 1992. How might emotions affect learning? In The Handbook of Emotion and Memory: Research and Theory, S.A. Christianson (ed.), 3–31. Hillsdale NJ: Lawrence Erlbaum Associates. Busel, V. (ed.). 2006. Great Explanatory Dictionary of Ukrainian Language. Kyiv: Perun. Christianson, S.A. (ed.). 1992. The Handbook of Emotion and Memory: Research and Theory. Hillsdale NJ: Lawrence Erlbaum Associates. Dzyuba, I. 2005. Internationalism or Russification. Kyiv: Kyiv Mohyla Academy. European Commission. 2001. Special Eurobarometer Survey 54. European Commission. 2006. Special Eurobarometer Survey 243/Wave 64.3 – TNS Opinion and Social. Isen, A.M. 1999. Positive affect. In The Handbook of Cognition and Emotion, T. Dalgleish & M.J. Power (eds), 521–539. New York NY: Wiley. Ivanishin, V. & Radevych-Vynnizkiy, Y. 1992. Language and Nation. Drogobych: Vidrodzhennya. Kolesnikov, V. 2003. Pereyaslav. Weekly 2000 34(184). Retrieved January 3, 2008, from http:// news2000.org.ua/print?a=%2Fpaper%2F24759 Kotsina, I. 2007. Sergey Taruta: Russian should be taught at schools not as a foreign language but on a par with Ukrainian. Facts 292: 3. Levkovska, N. 2006. Values dynamics of youth. Political Management 1(16): 85–93. Magochiy, P. 2007. History of Ukraine. Kyiv: Krytyka. Podobed, K. 2007. Opinion: Matter of time. Korrespondent 49. Retrieved January 3, 2008, from http://korrespondent.net/opinions/323227 Preobrazhenskaya, A. 2006. Russian is legalized. Retrieved November 24, 2007, from http:// www.partyofregions.org.ua/contrprop/resonance/444f3980d4979/ Shaizhin, A. 2006. Ukrainization a-la Valuev. Weekly 2000 26(324). Retrieved January 3, 2008, from http://news2000.org.ua/print?a=%2Fpaper%2F5478 Slezko, V. 2000. I would to another country like this... Kyivsky Telegraph 29. Retrieved January 3, 2008, from http://www.k-telegraph.kiev.ua/N29 Statistical Yearbook of Ukraine for 2003. 2004. State Committee of Statistics. Kyiv: Consultant. Statistical Yearbook of Ukraine for 2006. 2007. State Committee of Statistics. Kyiv: Consultant.
chapter 19
Proper names in the translation of The Lord of the Rings Vladimir Yepishev The Lord of the Rings is a multilingual trilogy which includes texts and lexical units of eight languages that are represented by means of the Roman alphabet and diacritics. This study scrutinizes the problem with translations of proper names. This includes an analysis of Tolkien’s languages and their phonetic system. The results obtained embrace sound value of graphemes as well as pronunciation rules and are used to compare proper names in the original with interpretations by Ukrainian and Russian translators, raising the question of phonetic correspondence in translation. The novel suffered much in translations because of various alternations and substitutions. Many proper names were misunderstood and therefore rendered incorrectly. Keywords: Sindarin, Quenya, Rohirric, Khuzdûl, translation, transliteration
1. Introduction The Lord of the Rings first reached the Russian reader in translations of Z. Bobyr’ and A. Gruzberg. Now there exist eleven Russian and three Ukrainian translations of the novel. Despite the impressive number of attempts to translate the trilogy made by both professional and amateur translators, the target text remains an adventure story only. Its linguistic features, such as Celtic-sounding of Sindarin are not preserved. The phonetics of the artificial languages, which were created by the author, undergo serious alternations, such as termination of diphthongs, sound substitution, consonant insertions, palatalization, etc. Consonantal and vowel changes upset the link between Tolkien’s transliteration and the graphic systems of Tengwar and Cirth, which he invented for his languages. For instance, according to these systems, d, t, dh and th should
Vladimir Yepishev
be represented by four different letters, which is hardly possible in the target text in Ukrainian or Russian. This makes Appendix E of The Lord of the Rings useless, as it describes things which are lost in translation. The Celtic-sounding Sindarin also serves the purpose of an intertextual link between the novel and Welsh stories of elves and fairies. Some of Sindarin words were originally taken by the author from the Welsh language, such as Sindarin duin “river” – dwyn “to carry”, Sindarin lin “pond” – llyn “lake”, etc. (Kist 2005). 2. The world of the novel The action of The Lord of the Rings takes place in an imaginary world of J.R.R. Tolkien, Middle-earth (derived from Old English Middangeard, as the professor commented in his letter to the New York Times: “Middle English midden-erd (or erthe), altered from Old English Middangeard, the name for the inhabited lands of Men “between the seas” . . .”). This is an “alternative” history of our world, where with men lived creatures that have long ago became myths: elves, orcs, trolls, ents, hobbits and many other beings. Middle-earth’s inhabitants have various languages and writing systems. Even birds and animals use languages. In The Hobbit (which is a preamble to The Lord of the Rings) the wargs were said to use a language of their own for communication (Tolkien 1974: 111). So were the eagles of the Misty Mountains and some birds that dwelled near the Lonely Mountain (Tolkien 1974: 118). The thrushes knew the language of Dale, which differed from The Common Speech so much, that the dwarves could not understand a word of it (Tolkien 1974: 268). Also the ravens were known to live long and master many languages (Tolkien 1974: 268). Even the stones of Middle-earth possessed a gift of speech, and the elves could hear them (Tolkien 1974: 301). Thus Middle-earth appeared before us as a world filled with magic, a world where the tale was alive and where the myths were reality. A fairy world, indeed. 3. Languages of the novel The novel encapsulates texts and lexical units of several languages, some of which are artificial languages created by the author himself. Among these are the languages of elves, dwarfs and orcs. The dominating language of the novel is Westron, which is said to have been completely substituted by the English language. The author plays the part of a reteller and a translator at the same time, rendering the story into English and finding a way to represent songs, poems and words, which are not of Westron origin. J.R.R. Tolkien did not want to flatten his
Chapter 19. Proper names in the translation of The Lord of the Rings
novel to English, and so he had to use new consonants, write pronunciation guides and write separate appendixes on “alien” languages. It must be mentioned that the novel is hard for the English reader, since some of the artificial languages have confusing consonants and vowels as well as unusual stress rules and diphthongs. What follows is a more precise analysis of the differences between languages. 3.1 Quenya Quenya – the language of the high elves, language of lore, an equivalent of Latin. It has five vowels: [a], [6], [i], [f], [u], represented by a, e, i, o and u correspondently. Quenya has fifteen consonants: [f], [g], [h], [j], [k], [l], [m], [n], [ŋ], [p], [r], [s], [t], [v], [w]. In Appendix E to The Lord of the Rings, J.R.R. Tolkien explains how to read words in elven origin, which he transcribed (Tolkien 1974: 946). It is interesting to know that the vowel system of Quenya resembles more the systems of Italian and Spanish (and even Ukrainian) than that of English. Quenya vowels are to be pronounced clearly, and often Tolkien would resort to putting diacresis over vowels to make English readers read them properly, or marking the dissyllabic pairs of vowels, for instance: Eressëa [6'r6s6"], Manwё ['m"nw6]. Besides short vowels, Quenya has long [a:], [6:], [i:], [f:], [u:], which Tolkien marks in his book with acute accent: á, é, í, ó and ú (Tolkien 1974: 946). Not all consonants in transliterated elven words preserve the phonetic value, which they have in the English language: some of them deserve a separate description; these are: c, h, f, k, ng, qu, r, s, th (Tolkien 1974: 948). The pronunciation of these vowels is scrutinized in Table 1. Table 1. Quenya-English transliteration and pronunciation Transliteration
Pronunciation
c In all elven words consonant c has a value of [k], and never of [s], as in some English words, like celedon, cell, etc. f Denotes [f] sound everywhere except at the end of words, where it is read as [v]. h Alone stands for English sound of [h]. k Same as c, but occurs only to denote words of non-elven origin, i.e., it can be observed in transliteration of Dwarvish words and that of the Black Speech. ng Stays for the [ŋg] sound, as in English words finger, linger, etc. qu Representation of [kw], as in English question. r In elven words represents trilled [r] in all positions, though in Dwarvish and the Black Speech it could stand for back or uvular r. s Voiceless English [s] in all positions. th In Quenya th has the value of [s].
Vladimir Yepishev
Double consonants in alien words represent either “double” or long consonants. Quenya has six diphthongs: ai, oi, ui and au, eu, iu. All these diphthongs belong to falling diphthongs and are intended to be pronounced respectively as [ai], [fi], [ui], [aʊ], [eu] and [iu] (Tolkien 1974: 949). 3.2 Sindarin The language of the Grey-elves includes five vowels (same as Quenya) plus a semi-vowel y, which indicates a sound of front German [y], as in the word Mücke (Tolkien 1974: 947). Long vowels in Sindarin transliteration are marked with acute accent or a circumflex, if a vowel stands in stressed monosyllables (Tolkien 1974: 947). The diphthongs are: ae, oi, ei, oe, ui, and au. Sometimes in accordance with the English custom the author uses aw instead of au at the end of words. The rules of diphthong pronunciation are the same as in Quenya. It is important to know that the vowel i standing in the beginning of the word before another vowel acquires the value of [j], as in the English word you, becoming a consonant. Concerning the consonants, Sindarin possesses several that do not occur in Quenya’s transliteration. These are: ch/kh, dh, gh, lh, ph, rh. Moreover, th consonant in Sindarin has the value of [θ]. Thus both elven languages possess the same grapheme, which has different phonetic meanings. In Sindarin, ng in the initial position denotes [ŋ] sound, as in the English word wing. Lh and rh represent voiceless l and r; gh is “back spirant” in the Black Speech. Some researchers believe lh denotes Welsh sound [S], though this theory is against the Appendix E of the novel. Some Sindarin consonants are given in Table 2 along with their pronunciation guides. Note that the table describes only those consonants that do not occur in Quenya’s. 3.3 Rohirric The language of Rohirrim had much in common with Anglo-Saxon (which is Old English). Anglo-Saxons became the prototype for Rohirrim, and their language inspired Rohirric, which led to Anglo-Saxons lexical units in Rohirric. No definite rules for pronunciation of Anglo-Saxon have been established, since it varied much across Britain, and also because scholars have not yet agreed about the pronunciation themselves. Table 2. Sindarin-English transliteration and pronunciation Transliteration
Pronunciation
ch/kh dh ph
Stands for [x] as in the German word Chemie. Denotes English sound of [ð]. Marks the [f] sound.
Chapter 19. Proper names in the translation of The Lord of the Rings
The language contained seven vowels: a, æ, e, i, o, u, y. Unlike modern English, y in Anglo-Saxon was always a vowel. The general pronunciation rules for vowels have been similar to the rules of most European languages, still different from Modern English (Allan 2006). The approximate phonetic values of the AngloSaxon vowels are given in Table 3. Table 3. Anglo-Saxon vowels pronunciation a e é i í o ó u ú æ y
[a:] [e] [ei] [i] [i:] [f] [6~] [u] [u:] [æ] [y]
Besides monophones there were six diphthongs in Anglo-Saxon: ea, éa, eo, éo, ie, and íe (Allan 2006). Most of the consonants sounded the same as in Modern English, but those that differed (Allan 2006) are given in Table 4. Table 4. Anglo-Saxon consonants pronunciation Letter
Position
Pronunciation
f
at the start or the end of word in the middle of word beside unvoiced consonant doubled
[f] [v] [f] [f]
s
at the start or the end of word in the middle of word beside unvoiced consonant doubled
[s] [z] [s] [s]
sc
usually
[w]
þ or ð
at the start or the end of word in the middle of word beside unvoiced consonant doubled
[θ] [ð] [θ] [θ]
h
at the start or the end of word in the middle of word
[h] [x]
c
in general before e, before i, after i
[k] [tw]
Vladimir Yepishev g
in general before e, before i, after i in the middle of word
[g] [j] [>]
cg
[dŠ]
ng
[ŋg]
The letters þ and ð were interchangeable. Doubled letters denoted longer sounds, as in Italian and Finnish. All letters were pronounced. Note that some sources give different information on Old English pronunciation rules, for instance è is suggested to be pronounced as [æ]. Again according to other sources ch is supposed to have had the value of [kh] and sc is [sk] (Stoudenets’ 1998: 35), while the vales of [tw] and [w] are attributed to the Middle English period (Stoudenets’ 1998: 30). 3.4 The Black Speech This is the speech of Sauron’s servants. Any empire needs an official language, and the Black Speech became the official language of the domain of Mordor – the land ruled by Sauron. It possesses the plosives b, g, d, p, t, k, the spirants th, gh (and possibly f and kh, attested in Orc-names only), the lateral l, the vibrant r, the nasals m, n, and the sibilants s, z, sh (Fauskanger 1999). Due to lack of information on the Black Speech, this list might not be full. There seem to be only four vowels in the speech of Sauron’s servants, which are a, i, o, u; note that o is stated by Tolkien to be rare. Long vowels are attested, as in Sindarin transliteration (Tolkien 1974: 950). Sadly, J.R.R. Tolkien did not leave much information on the Black Speech of Mordor and its pronunciation. Orcs were said to adopt foreign languages and change them to their own liking, but they created brutal jargons, insufficient even for their own small needs (Tolkien 1974: 937). Orcs had many dialects, and sometimes the difference between them was so great that members of one tribe could not use their language for inter-tribal communication. 3.5 Khuzdûl Khuzdûl is a secret language of the dwarven race. Not much can be said about it, and not much is known about it. Digging through all books on Middle-earth, only a few words of genuine Dwarvish can be found. However, there are some points which are to be highlighted: unlike in transliterated words of elven origin, aspirated th and kh do not represent a single sound of [θ] or [x], but [th] and [kh] respectively, as in the English words backhand and outhouse (Tolkien 1974: 937).
Chapter 19. Proper names in the translation of The Lord of the Rings
4. Transcribing into Ukrainian: Problems and solutions Transcription into Ukrainian is easier than into English, for Ukrainian possesses more means for rendering sounds than the English language. Nevertheless, there are some issues: representing the [h] sound, rendering [ŋ] sound, [ð] and [θ] sounds transfer. Traditionally translators prefer transcription of elven words instead of their transliteration. Unfortunately, the Ukrainian language has no consonants to denote [ŋ], [ð] and [θ] sounds, so translators have been forced to follow the tradition of substituting them with other letters that have different phonetic values (Korunets’ 2001: 92–93). This substitution however, may often lead to serious misunderstandings. For instance, there are two words: Tauron and Thauron, which will be both rendered as Таурон [tauron] (since voiceless th is often substituted with t by Ukrainian translators), unless they both occur in the text. The problem is that these two names belong to two different characters, who are mortal enemies: a protagonist Tauron, whose name is formed from taur, which means “forest” and Thauron, whose name comes from thaur – “abominable, abhorrent”. In translations with substitutions they will merge into a single person, and if the first one is a forester and the second one is abhorrent, then together they form someone who is an “abhorrester”. One of the possible ways out might be lo leave all non-English words as they are given in the text – a method implied in translating into Ukrainian The Legend of Thyl Ulenspiegel and Lamme Goedzak in 1980s. The novel was written by a famous Flemmish writer Ch. De Coster. It was written in French but contained many Dutch words which were represented in their original form. When in the 1980s an attempt to translate the book into Ukrainian was made, a translator from the French language faced the baffling task of rendering words he could hardly comprehend. Luckily the book had a glossary made by the author in which he provided a descriptive translation of the Dutch words into French. The solution was the following: the translator rendered all French units and left all the Dutch words unaltered, as they were in the original. Thus he received a text containing Ukrainian and Dutch with a Dutch-Ukrainian glossary.
5. Alternations in proper names rendering Names prove to be the most confusing for translators to convey. One difficulty is caused by J.R.R. Tolkien’s use of consonants and characters, which do not occur in English (dh, lh); as another by his combinations of characters and letters that are valued differently from the English (c, ch). To support the argument, several
Vladimir Yepishev
elven names are presented in their original transliteration by J.R.R. Tolkien and translations made by Ukrainian and Russian translators. All phonetic shifts in the translation of proper names can be classified into two major groups: alternations made accidentally and those made deliberately. To the accidental group belong proper names in so called “underground” translations. Despite the fact that Tolkien’s novel was translated in the late 70s, it was not published officially. The Lord of the Rings was considered to be an antiSoviet novel. Maybe Tolkien had not intended it to be so, but the Soviet editors did not publish his novel. Yet the translations were not burned, and they spread: written by hand and told by mouth, the tale of the Ring traveled across the ussr. Naturally the first translators of The Lord of the Rings had no access to other Tolkien’s books and articles about the languages of the novel. Thus the distortion of proper names in first translations of the trilogy is justified by translators’ unawareness of the multilinguality of the novel. The deliberate distortions and alternations took place in other translations that followed. They are not owing to mistakes of young translators, but to the actions chosen by venerable authorities. All deliberate and accidental alternations can be subdivided into the following subgroups: ☐ alternations caused by translator’s unawareness of the text’s multilinguality; ☐ distortions resulting from the translator’s will to drop certain sounds and substitute them; ☐ alternations caused by the will to assimilate proper names to the language of translation; ☐ alternations made when the word is considered to be rude; ☐ distortions by translation of what should remain unchanged; ☐ transliteration of proper names which should be translated; 5.1 Distortions in proper names transcription Unfortunately distortions in the transcription of proper names are the most frequent and no translation of the novel has ever been made without a certain phonological twist of names. Examples of this kind of misunderstanding are found even in the most up-to-date Ukrainian translation by O. Feshovets’, which is said to have been performed following the author’s note on pronunciation. 5.1.1 Lúthien ['lu:θi6n] This is an elven name of Sindarin origin, therefore th here represents unvoiced interdental fricative [θ]. A rough conveying of this name would be Лутіен for Ukrainian and Лутиэн for Russian. Table 5 provides variants given by translators. It is interesting to note that some of the distortions are the result of “free translation”. Lúthien was very beautiful and maybe this was the reason why N. Grigorieva
Chapter 19. Proper names in the translation of The Lord of the Rings
Table 5. Lúthien transcription into Cyrillic Variant
Translator/s From English into Russian
Лучиэнь [Álutwi6nj] Лютиен [Áljutij6n] Лютиэн [Áljuti6n] Лютиэнь [Áljutij6nj]
N. Grigorieva and V. Grushetskii Z. Bobyr’ A. Gruzberg V. Volkovskii, D. Afinogenov and V. Tikhomirov A. Kistiakovskii and V. Muraviov
From English into Ukrainian Лючіень А. Nemirova [Áljutwi7nj]
and V. Grushetskii gave her a new name, which contains the stem of Russian word “луч”, (a ray). It can be also assumed that translations containing “лют” might be attributed somehow to the name of a flower in Russian: “лютик” (a buttercup). Substituting interdental [θ] with other consonants ([t], [s] or [f]) became a tradition in transliteration (Korunets’ 2001: 98). Indeed Ukrainian and Russian do not possess a grapheme for this voiceless dental fricative: fita is used no more. The question is whether such a transformation is a distortion, and the answer is affirmative. 5.1.2 Grishnákh [Ágriwna:x] This is the name of the orc commander, who was sent from Mordor with troops to seize the Ringbearer. The name is of the Black Speech origin. As we can see from the name Grishnákh, it contains sh sound, which does not occur in elven languages and might sound ill to elves; in addition, we may assume that the vowel i could be pronounced in a different way, for instance, as Ukrainian и, which would probably grate upon the elven ears as well. Orcs often growled and grunted, and it is more likely that they preferred back vowels (to which Ukrainian и belongs) instead of the front ones. Note that kh denotes a single sound of [x]. According to the sounding of the name, its faithful translation into Ukrainian is Ґришнах. In Ukrainian and Russian translations it has been rendered as it is shown in Table 6. The cardinal error is the lack of either k or x sound in the target language variants. A. Nemirova’s variant of the name contains the o sound, which is known to be extremely rare in the Black Speech: there are only few words that contain the sound o in the orcs’ speech (Fauskanger 1999).
Vladimir Yepishev
Table 6. Grishnákh transcription into Cyrillic Variant
Translator/s
From English into Russian
Горшнак [gfrwnak] Гришнак [griwnak] Гришнакх [griwnakx] Грышнак [grGwnak]
A. Nemirova N. Grigorieva and V. Grushetskii Z. Bobyr’ V. Volkovskii, D. Afinogenov and V. Tihomirov A. Kistiakovskii and V. Muravyov
From English into Ukrainian Горшнак A. Nemirova [Bfrwnak]
5.1.3 Lugbúrz [Álugbu:rz] This was the name of Barad-dûr, the tower of Sauron, in the language of orcs. It is not a difficult name to render. The right transcription of it would be Лугбурз [Álugburz] for Russian and Луґбурз [Álugburz] for Ukrainian. Somehow it seems unsuitable for the translators, as Table 7 shows. As it is seen from the table above, some translators perceive the Black Speech as German, and some choose to invent their own variants for the word instead of transcribing it. Table 7. Lugbúrz transcription into Cyrillic Variant
Translator/s From English into Russian
Лугбурц [lugburu] Лагбур [lagbur] Лугбур [lugbur]
A. Gruzberg A. Nemirova N. Grigorieva and V. Grushetskii V.A. Matorina
From English Into Ukrainian Горбурц A. Nemirova [gorburu] O. Feshovets’ Луґбурц [lugburu]
Chapter 19. Proper names in the translation of The Lord of the Rings
5.1.4 Eowyn [ejfwyn] “Thus Aragorn for the first time in the full light of day beheld Éowyn, Lady of Rohan, and thought her fair, fair and cold, like a morning of pale spring that is not yet come (Tolkien 1974: 720). to womanhood” Aragorn saw Éowyn, the niece of Théoden, king of Rohan. The language of Rohan was Rohirric, though the Men of the Mark spoke the common language as well. As it was mentioned above, Rohirric was inspired by Anglo-Saxon, which J.R.R. Tolkien knew well. The usage of Anglo-Saxon words in Rohirric proves the application of the same pronunciation rules as the Anglo-Saxon language: é represents a sound as in the English word bay; while y denotes that of German für. Guided by these rules, a translator would render the name as Ейовюн into Ukrainian. Russian and Ukrainian variants of this name can be found in Table 8. There are also such variations as Эйовин [6jovin] and Эовейн [6ov6jn], unfortunately erroneous as well. O. Feshovets’ and A. Nemirova changed this name in their translations to Éowyn Еовіна [7f‚ina].
6. Conclusion Sadly all the multilinguality of The Lord of the Rings gets flattened in the process of translation. Comprised of several languages with separate phonology and writing systems, the novel loses some of its linguistic features. Ukrainian and Russian readers do not receive the enchanting Celtic sounds of Quenya or Sindarin, they do not feel the difference between the language of dwarfs and the harsh languages of orcs. All sound the same, being limited to the phonetics of the target language. Table 8. Éowyn transcription into Cyrillic Variant
Translator/s
From English into Russian
Йовин [jovin] Эовин [6ovin]
N. Grigorieva and V. Grushetskii Z. Bobyr’ A. Kistiakovskii and V. Muraviov
From English Into Ukrainian Еовіна A. Nemirova O. Feshovets’ [7f‚ina]
Vladimir Yepishev
Sound substitution leads to character distortion. The names of characters influence the way we perceive them: when names are changed, the perception is changed: Old Norse Þórr does sound like a roll of thunder, while Russian variant [tor] is just a word with a dim meaning.
References Allan, S. 2006. Pronunciation of Old English. Retrieved November 30, 2007, from http://www. beowulftranslations.net/pronunciation.shtml Fauskanger, H.K. 1999. Orkish and the Black Speech. Retrieved November 30, 2007, from http:// www.uib.no/people/hnohf/orkish.htm Kist, O. 2005. Elves, Sindarin and Quenya. Retrieved January 9, 2008, from http://www.tolkienics. com/tolkien/sindarin.htm Korunets’, І.V. 2001. Theory and Practice of Translation. Vinnytsia: Nova Knyha. Lobdell, J.C. 2003. A Tolkien Compass. Chicago IL: Open Court Books. Stoudenets’, G.I. 1998. The History of the English Language in Tables. Kyiv: KSLU. Tolkien, C. & Tolkien, J.R.R. 2000. The Letters of J.R.R. Tolkien. New York NY: Houghton Mifflin. Tolkien, J.R.R. 1964. Tree and Leaf. London: Unwin Books. Tolkien, J.R.R. 1974. The Hobbit (Collector’s Edition). Boston MA: Houghton Mifflin. Tolkien, J.R.R. 1974. The Lord of the Rings (Collector’s Edition). Boston MA: Houghton Mifflin. Tyler, J.E.A. 2004. The Complete Tolkien Companion. New York NY: Thomas Dunne Books.
chapter 20
Threat and geographical distance The case of north Korea Jan Prasil, Maria Dudusova & Jan Auracher North Korea is considered by westerners to be a dangerous and unpredictable country and its leader, Kim Jong II, a mad dictator. After its nuclear test in October 2006, nervousness reached a new critical point. Yet first impressions gathered after this test showed that perceptions of this “rogue state” range widely among observers in the West and their Asian counterparts. The present research specifically assesses the fear of subjects from four countries, when confronted with the official press release after the event. Our hypotheses were that the perceived fear of North Korea would differ between subject groups of each country, and that it would increase in correlation to geographical distance. The results confirm the first hypothesis. The second, however, could not be verified. Keywords: perception, fear, credibility, enemy, nuclear test, North Korea, geographical distance
1. Political background of the research – what happened before? The time when the northern regions of Korea were referred to as “The Land of the Morning Calm” is long over. For more than fifty years the United States has been at loggerheads with The Democratic People’s Republic of Korea. Under George W. Bush’s term these political interferences intensified. In 2002 North Korea was placed in the “axis of evil”, and the country was attacked verbally. For example, Colin Powell, the U.S. Secretary of State, described it as “a dark, cold and hungry hell” (Sommer 2003). In this context, U.S. administration rhetoric started to use the term “rogue state”. Both the country and its dictator matched the classical concept of an enemy. That a war of aggression could be launched by North Korea against Seoul or Tokyo, and carried out with weapons of mass destruction without regard for its
Jan Prasil, Maria Dudusova & Jan Auracher
own losses, was consequently considered entirely possible in the Western world. On the other hand, the fear of a potential attack was mutual. Ri Chan Bok, General of the People’s Army, declared that North Koreans were ready to defend themselves anytime (Rather 2006). These words were repeated again when North Korea claimed to have successfully tested a nuclear bomb on October 9, 2006. For one whole week there were speculations around the world about the truthfulness of the nuclear test. The media asked: “Was it really a bomb?” The October 10 issue of the German newspaper ‘Die Welt’, for example, published critical comments from experts, who asked whether the detonation stemmed from tnt (Bodderas 2006). Given the weakness of the test, nobody was sure if it was a conventional or indeed a nuclear explosion. “More likely, scientists say, the test did not take place” (Spiller 2006). Nevertheless, many US media presented the test as a fact already on October 9: “North Korea (…) successfully tested a nuclear bomb this morning at a site near the city of Kilju” (Powell 2006a). On October 16, the U.S. Government confirmed that the test was indeed nuclear. Though doubts about the truthfulness of North Korea’s nuclear claims remained, the public reaction resulted in a state close of hysteria. The international community then confirmed its suspicion that the “rogue state” was indeed a dangerous nuclear threat, thereby legitimating the international sanctions against it. At this time, nervousness in the USA reached a new critical point. Newspapers issued headlines such as “When Outlaws Get The Bomb” (Powell 2006b), “A Nuclear Nightmare Comes True” (Powell 2006a), “The Axis Bomb” (New York Sun 2006). Also in other countries mass media created a picture of a serious threat. In Germany, the perception of North Korean borders changed and became very near, offering a geographical risk: “Very close to North Korea […] An unscrupulous country, that has nuclear weapons and needs money, exists in times in which the Islamic terrorists want to bring the greatest possible damage over the West, even not on the other side of the world, but just around the corner.” (Mayntz 2006) In Japan, the Prime Minister described the nuclear weapons testing as a “serious threat [that] would transform in a major way the security environment of North East Asia.” (Pilling 2006a). Yukio Okamoto, a security expert and a former special adviser on foreign policy to Junichiro Koizumi, Mr Abe’s predecessor, said: “In Japan, because of this threat, which people are feeling for the first time in the post-war period, there is going to be a massive outcry to beef up Japan’s security.” (Pilling 2006b)
2. The research plan Given this situation, we wanted to determine the extent to which the perception of this threat differed between individuals of different cultural backgrounds.
Chapter 20. Threat and geographical distance
To this end we surveyed subjects from four different countries, Japan, Poland, Germany, and the USA, about their fear of the “North Korean bomb”. The aim of the research was to test whether geographical distance affects the degree of perceived danger. We believed that cultural and historical proximity exerts an influence on the way information about countries is processed. Accordingly, it was assumed that the North Korean regime would be perceived in the light of the experience of the respondents. That is, we hypothesized that the lack of information in countries situated at the greatest geographical distance (Germany, Poland, the USA) would be likely to cause a feeling of uncertainty with respect to North Korea. Living in the neighborhood to North Korea, the average Japanese has rather different experiences from people in Europe or North America. For one, Japan has quite a lot of North Korean immigrants, which enables Japanese to get information first hand. Moreover, in Japanese media people regularly can read about incidents with North Korea, for example when smugglers try to bring false money over the Japanese Sea or when North Korea launches missiles towards Japan. Despite this daily awareness of the actual relation with North Korea, Japan shares a history of more than a thousand years with its neighbor. Japanese grow up with stories and images that are passed on within their society. The situation in North America and in Europe is certainly quite different. People there get less information and mostly from actual issues in mass media. The average European or North American usually hears from North Korea only in the context of press releases such as those quoted above. The level of information in the three tested areas, hence, is completely different. Therefore, we assumed subjects from neighboring countries would assess the danger coming from North Korea in the light of their experience less emotionally than subjects from distant countries. Thus, our hypotheses were the following: 1. Subjects from different countries present different levels of fear. 2. Fear increases in correlation to geographical distance.
3. Materials The following document was selected to test the differences in perception between the three western countries and Japan. The document was an official statement of North Korea’s government. Pyongyang, October 9 (kcna) – The Korean Central News Agency released the following report: The field of scientific research in the dprk successfully conducted an underground nuclear test under secure conditions on October 9, Juche 95 (2006) at a stirring time when all the people of the country are making a great leap forward in the building of a great prosperous powerful socialist nation.
Jan Prasil, Maria Dudusova & Jan Auracher
It has been confirmed that there was no such danger as radioactive emission in the course of the nuclear test as it was carried out under a scientific consideration and careful calculation. The nuclear test was conducted with indigenous wisdom and technology 100 percent. It marks a historic event as it greatly encouraged and pleased the kpa and people that have wished to have powerful self-reliant defense capability. It will contribute to defending the peace and stability on the Korean Peninsula and in the area around it. (Official press release of the Korean Central News Agency)
To test our hypotheses (see Section 2), 98 participants from four countries, of three different geographical regions (Japan – East Asia, Germany and Poland – Europe, and the USA – North America) were invited to participate. All participants were citizens in their countries and received the text in their mother language. We used the translations that were published in the national media. Subjects were asked to read the statement carefully and respond to a questionnaire. The questionnaire consisted of six statements, which were to be assessed on a 7-point Likert scale according to the subject’s approval (1 = “not at all” to 7 = “completely”). The statements referred to two dimensions concerning, on the one hand, the perceived threat of North Korea’s nuclear technology and, on the other hand, the credibility of North Korean regime. Each dimension (credibility and fear) included three statements, which had to be assessed separately.
Credibility: How safe do you consider Korea’s nuclear technology to be? How likely do you think it is that North Korea can afford to develop nuclear program? Would you describe North Korea as a country with the fifth biggest army, as a superpower?
Fear: North Korea is striving for the hegemony in the region. How likely do you think that is? North Korea will use its nuclear capability for hostile purposes. To what extent do you agree? Do you feel threatened by North Korea?
. KCNA: http://www.kcna.co.jp/item/2006/200610/news10/10.htm
Chapter 20. Threat and geographical distance
4. Subjects Most of the 98 participants were university students of various educational institutions, mainly social studies students aged 19 to 27. As regards gender, 59 of them were female and 39 male. In relation to nationality, 25 subjects were from Germany, 21 from the USA, 19 from Poland and 33 from Japan. Table 1. Subjects by nationality, gender and age Nationality Germany USA Poland Japan
Female
Male
Average age
14 11 11 23
11 10 8 10
27 22 25 20
5. Results 5.1 Hypothesis 1 The six statements of the questionnaire were divided into two dimensions. The first examined credibility, the second, fear. The correlation of the items within each dimension was tested by a factor analysis with Rotated Component Matrix. Differences were observed on both dimensions. Table 2 shows the results: Table 2. Credibility and fear
Germany
Poland
USA
Japan
Standard Standard Standard Standard Mean deviation Mean deviation Mean deviation Mean deviation Credibility 2.34 Fear 4.0
1.43 1.44
3.50 3.89
1.29 1.57
1.11 1.28
1.40 1.18
2.31 5.35
1.75 1.38
In the eyes of the German and Japan subjects credibility was nearly the same. Polish subjects exhibited a large difference (1.1) compared to its neighboring country, Germany. However, in the dimension of fear, the results of these two countries were nearly identical. The Polish participants trusted the credibility of North Korean regime and were afraid of it. The Americans did not trust the regime and were not afraid of its nuclear power. Only Germans and the Japanese did not trust the capabilities of the “rogue” regime, but were afraid of it. The difference between the lowest degree of fear and the highest that is between American and
Jan Prasil, Maria Dudusova & Jan Auracher
Japanese subjects, was substantial (4.07). The results obtained from Japanese and American participants deviated clearly and exhibited extreme values. The German and Polish results were average. The highest degree of credibility was expressed by the Polish participants (Mean 3.5; Standard Deviation 1.29). They were followed by the German participants (Mean 2.34; Standard Deviation 1.43) and the Japanese (Mean 2.31; Standard Deviation 1.75) subjects. The lowest credibility of the North Korean regime was given by the participants from the USA (Mean 1.11; Standard Deviation 1.4). anova revealed a highly significant influence of the variable nation on both credibility (F = 5.394, p < 0.001) and fear (F = 13.923, p < 0.001). Significant influence of gender or age on credibility could not be observed. However, there were significant differences concerning fear depending on age (F = 5.399, p < 0.01) and gender (F = 3.225, p < 0.05). These results confirmed that there are significant differences between the countries examined. 5.2 Hypothesis 2 Our results did not confirm the hypothesis that fear increases in correlation to geographical distance. In the geographically nearest country, Japan, there seems to be, according to our results, more demonstration of fear. The highest results for fear perception were given by the Japanese subjects (Mean 5.37; Standard Deviation 1.38), followed by the German (Mean 4.0; Standard Deviation 1.44) and Polish (Mean 3.89; Standard Deviation 1.57) subjects. The lowest rate for fear was shown in the results of the American participants (Mean 1.28; Standard Deviation 1.18). We could not confirm the hypothesis that fear is reduced by the access to more and more detailed information in the vicinity of North Korea. The results show that there is a large (4.09) difference between the level of fear in Japan and the USA. On the other hand, we can state that the fear levels in the neighboring countries of Germany and Poland are nearly the same.
6. Conclusion This study confirmed that there are similarities in the attitude towards North Korea among participants within one country and that these cultural attitudes do significantly differ according to geographical distance. While the answers from the Japanese and the American subjects took the extreme positions for both dimensions (credibility and fear), the two European countries tended more towards neutral answers. The participants from Japan felt most threatened, followed by the participants from Germany and Poland. Between the German and the Polish group there are only minimal differences. It is therefore plausible to assume that there may be other cultural factors which influence people’s opinions and fears.
Chapter 20. Threat and geographical distance
In contrast to our second hypothesis, however, the participants from the United States proved to exhibit the lowest degree of fear. At the same time, they also assigned the lowest level of credibility. This is in opposition to the official argumentations of the government and to the villainous image of North Korea used in the mass media. Interesting is the result that German and Japanese subjects did not trust the capabilities of the “rogue” regime, but were afraid of it. The numbers demonstrate a conspicuous discrepancy between the credibility and the fear. With regard to the fact that comments by the government and the media reports in Germany and Poland strongly correspond to the American ones, the differences between the results of the subjects from these countries are considerable. Based on the results of this study, we can assume that there is fear of North Korea in Germany and Poland in spite of the fact that the potential threat would in all likelihood be directed at the USA or Northeast Asia.
References Bodderas, E. 2006. War es wirklich die Bombe? Welt Online October 10: http://www.welt.de/ print-welt/article158482/War_es_wirklich_die_Bombe.html Mayntz, G. 2006. Nordkorea ganz nah. Rheinische Post October 20: http://www.presseportal. de/pm/30621/889567/rheinische_post New York Sun. 2006. Editorial, October 10: http://www.nysun.com/article/41218 Pilling D. 2006a. Abe calls test “absolutely unacceptable”. Financial Times October 09: http:// search.ft.com/ftArticle?queryText=serious+threat+%5Bthat%5D+would+transform+in+a +major+way+&y=0&aje=true&x=0&id=061009004219&ct=0&nclick_check=1 Pilling, D. 2006b. Pyongyang test sparks Abe’s first crisis. Financial Times October 10: http:// search.ft.com/ftArticle?queryText=“to%20be%20a%20massive%20outcry”%20&y=0&aje =false&x=0&id=061010000976&ct=0 Powell, B. 2006a. A Nuclear Nightmare Comes True. Time October 09: http://www.time.com/ time/world/article/0,8599,1544017,00.html Powell, B. 2006b. When Outlaws Get the Bomb. Time October 15: http://www.time.com/time/ magazine/article/0,9171,1546342,00.html Rather, D. 2006. North Korea: The Hermit Kingdom. CBS News Jan. 16: http://www.cbsnews. com/stories/2006/01/12/60minutes/main1203973.shtml Sommer, Th. 2003. Eine Hölle, kalt und dunkel. Die Zeit February 20: http://www.zeit.de/2003/ 09/Nordkorea Spiller, P. 2006. N Korea test – failure or fake? BBC News 16 October: http://news.bbc.co.uk/1/ hi/world/asia-pacific/6040494.stm
chapter 21
The apology of popular fiction Everyday uses of literature in Poland Maciej Maryl Popular fiction is expelled from literary canons and is usually treated as a form of ludic or emotional relaxation and an escape from daily problems (Nell 1988; Radway 1984). Drawing on the results of my study on Polish readers, I claim that reading popular fiction, as an everyday-life activity, is much more than a simple leisurely activity, and plays a vital role in the way readers cope with their surrounding environment. Employing the framework of phenomenological sociology I argue that reading popular fiction can be described in terms of a rational action undertaken in order to achieve specific pragmatic goals, such as emotional tuning or search for subjectively important information. Keywords: phenomenological sociology, life world, reading, popular fiction, canon, Poland
1. Introduction Why are some books perceived to be better than others? It is customary for works considered memorable for a certain culture to be listed in various canons. A knowledge of the canon marks one’s cultural competence. Sometimes, however, the other functions of reading, or its distinctive role in readers’ everyday lives are largely overlooked, and literature is perceived only through its social role, as the recent changes in the canon of required reading in Poland attests. The revision of the canon was proposed by Roman Giertych, deputy Prime Minister and Minister of Education in Jarosław Kaczyński’s conservative Cabinet from May 2006 to August 2007. The canon he planned to revise is very important in Poland, as it contains readings that are required at matura – a high school ending examination. On May 2007 Minister Giertych proposed to cross out Witold
Maciej Maryl
Gombrowicz, a great writer and critic of Polish nationalism and for non-Polish authors he left no room for Kafka or Dostoyevsky. Instead, Giertych added to the list two books by Jan Dobraczyński, a Catholic writer. The other author he added was John Paul II and two of his works: a book of essays (full text) and his biography (Giertych 2007). The canon was reshaped in order to convey the ‘right’ ideology, namely conservatism, Catholicism, and a specific version of blind patriotism, rooted in the hard times of the 18th and 19th century and criticized ever since. The debate provoked by Giertych’s decision, which swept through the Polish public sphere in the following months, did not undermine the basic concept of the literary canon, understood as a codification of books, which present ideas significant and fundamental to a certain community. The controversy was then not solely about the books, but about their ideological perspective. The discussion and its outcome (rejection of Giertych’s proposal by the Prime Minister) was one of the factors leading to the collapse of the ruling coalition and early elections. Although this is an extreme example of imposing ideology on the canon of required reading, it tells us something about the ways we perceive the canon in our cultures. The concept of the canon as a pillar of society draws heavily on the romantic understanding of literature. In 1765, Samuel Johnson pictured literature as an expression of general nature. Yet, in 1773 Johan Gotfried von Herder introduced the notion of literature as the extract of national spirit, which influenced the philosophy of romanticism. Due to political changes in the19th century and the emergence of the concept of nation (Anderson 1983), literature started to be conceived in a rather different perspective, as a nation-building factor, and was taught that way at universities (Hillis-Miller 2002: 2–8). It is through literature that the members of a community learn where they belong, and construct their social, cultural and national identities. Drawing from Fish’s (2004) concept of interpretive community, we may say that the canon is what the members have agreed upon to be valuable for the community itself. Community members, we can say from this perspective, choose certain books and agree upon the way they should be interpreted. Paulson (1997) provides a definition which captures this notion accurately. According to him, the canon is “the nebulous collection of works that at a given time are seen as enduring and worthy of study, and with which serious aspirants to membership in literary communities must be acquainted” (1997: 227). In the preface to The Western Canon (1994), Harold Bloom characterizes the discussion of the canon, as a debate between the right-wing defenders of the Canon, who wish to preserve it for its supposed (and nonexistent) moral values, and the academic-journalistic network I have dubbed the School of Resentment, who wish to overthrow the
Chapter 21. The apology of popular fiction: Everyday uses of literature in Poland
Canon in order to advance their supposed (and nonexistent) programs for social change (1994: 4)
This notion captures accurately the approach to the reading lists discussed before. Bloom, however, distinguishes another type of the canon, which he defines as “the relation of an individual reader and writer to what has been preserved out of what has been written” (ibid.: 17). Bloom seems to believe in the canon as a rather universal set, independent of social or political currents, since its principles of selectivity are “founded upon severely artistic criteria” (ibid.: 23). Hence, he distances himself from debates over the literary canon, claiming there are some works in which aesthetic quality is indisputable and universal. The difference between those approaches does not only concern the criteria of selection (memorable ideology vs. literary values), but also the role literature plays in readers’ lives. Bloom questions social uses of literature when he claims that “reading deeply in the Canon will not make one a better or a worse person, a more useful or more harmful citizen” (ibid.: 30). Instead he adds: “All that the Western Canon can bring one is the proper use of one’s own solitude, that solitude whose final form is one’s confrontation with one’s own mortality” (ibid.). In other words, what is the most valuable here is the unique encounter between the reader and the text, not the memorable meaning the text conveys. Yet, the question remains, how should we treat popular literature that is excluded from literary canons of both types, since it seems not to convey any values fundamental for a community, nor it is of exceptional literary quality. Critics often overlook the positive aspects of popular literature, treating it only as a part of a mighty machinery of mass culture. They conceive of it as “stories of adventure and excitement which satisfy desires for variety frustrated by urban living” (Kernan 1973: 40). A similar, but not so critical approach, may be found in studies into popular readership (Nell 1988; Radway 1984). Those researchers view the act of reading popular fiction as a form of ludic or emotional relaxation and an escape from daily problems. Although I do agree with such a perspective, I claim that reading popular fiction, as an everyday-life activity, means much more than a simple leisurely activity, and plays a vital role in the way readers cope with their surrounding environment. In this paper I show how readers use popular literature as a tool for achieving various goals. The experience of popular fiction, I will claim, could be as much rewarding for an individual as the exploration of the canon. Although such approach was presented before (eg., in Radway 1984) I would like to discuss everyday uses of literature in Poland, in the light of institutional approach towards reading, mentioned in the opening paragraphs of this paper. I reckon that the framework of phenomenological sociology applied in this paper may provide interesting insights on this field.
Maciej Maryl
2. The canon and everyday uses of literature Popular fiction is non-canonical, what evokes a specific feeling of guilt in readers. In the course of my study (Maryl 2007), I undertook a participant observation in a public library. When my researcher’s identity was revealed to a lady in her fifties, who was putting a number of romances on the counter, she cried: “These are not my usual readings, I’ve read the classical books, but I need something lighter right now”. This sense of guilt stems from the awareness of disobedience to cultural norms of reading. The study (ibid.) was mostly based on interviews with employees of public libraries. Librarians would rather see their clients reading canonical works, but admit that common readers tend to prefer popular fiction to canonical readings. From the readers’ perspective the canon is something culturally approved and desired, but at the same time considered difficult and boring, and as such, dismissed. Moreover, common readers feel the cultural pressure on their choices and they ‘know’ that they ‘should’ read the classical and acclaimed fiction, but they do not. As one of the librarians put it – they suffer from the “unbearable lightness of their readings”. Nell traces the roots of this feeling of guilt to Protestant ethics, which entails “that pleasure must be earned, and that the effortlessness of ludic reading makes its pleasures as hollow as the euphoria of the junkie or the orgasm of the masturbator” (1988: 32). Nell claims that this dissonant, caused by reading literature acknowledged as simplistic, can be resolved in two ways: “One is to acknowledge that I do in fact read trash, but that I have a moral license to do so; the other is to argue that while many people read trash […] my own reading matter is clearly not trash” (ibid.: 44). As Radway observes in her study on female romance readers “Guilt arises as a result of the readers’ own uneasiness about indulging in such an obviously pleasurable experience” (1984: 106). Both Radway and Nell agree that readers resolve this dissonant by claiming the information value of the books they read (Radway 1984: 112–114; Nell 1988: 44). Women in Radway’s study believed “very strongly that romance reading is worthwhile because the stories provide pleasure while the activity of reading challenges them to learn new words and information about a world they find intriguing and all too distant” (ibid.: 116). Yet, I believe that this informational function of popular readings shall be considered as something more than a simple justification of readers’ choices. I claim that readers use books in order to fulfil their own goals, which are not entirely of a . All quotes from the study in Maryl 2007 I translated from Polish.
Chapter 21. The apology of popular fiction: Everyday uses of literature in Poland
ludic or emotional nature. In order to support this claim I will draw on the results of my study in which I investigated the role of literature in readers’ everyday lives (Maryl 2007). Employing the framework of phenomenological sociology, I wanted to investigate how readers take the advantage of literature in coping with daily life. This approach is different from the one discussed by Paulson (1997), who concentrated on practical aspects of literacy and literature’s contribution to improvement of one’s professional skills (e.g., decision-making). In my study I am concentrating on less abstract goals determined by one’s everyday-life. Later in this chapter I will claim that some everyday reading practices are not a simple leisure, being more complex and having specific pragmatic goals. I employed the theoretical framework of phenomenological sociology, focusing on the life world and the actor’s practical interest. Life world, a concept based on Edmund Husserl’s notion of Lebenswelt (Husserl 1970), is a phenomenological construct through which we experience reality. Life world, in general, is the world as we see it and experience it, a sort of a mental filter, through which we select information, being guided by our practical interest (Schutz & Luckmann 1973). As Schutz (1964) claims, our life worlds are divided into several spheres of importance, which are put into hierarchical order according to our practical interest (from the most important sphere, to the least important one). Robert Gorman provides a good summary of Schutz’s argument: we are all unique actors, each a product of a biographically determined situation, belonging to only one person. Meaning and knowledge, those factors determining how we define our situations and act, are constituted subjectively through our perceiving and experiencing the world (1975: 6).
Every action undertaken by actors has its ‘in-order-to’ motive: a “future desired state of affairs as anticipated when action is begun” (ibid.: 4). All such motives are subject to the actors’ life world: “each actor experiences and defines his situation and chooses his projects in the context of his own unique, subjective existence” (ibid.: 3). In other words, Schutz claims that every action has its purpose, which is dictated by one’s own biography. I will claim that reading popular fiction is such an activity, through which readers achieve individually determined goals and enrich their life worlds. The structure of the life world, a biographically determined situation, conditions a reader’s response. The goals achieved through reading, the experience gained in the fictional world, constitute a part of the life world. This active search for information and emotions is deeply rooted in one’s everyday experience. The patterns of readers’ behavior, presented below, outline the importance of their everyday motives. That is the main difference between socio-phenomenological
Maciej Maryl
inquiry and the psychological approach taken by Nell, who measured the ‘interest in a book’ by “asking readers to decide which book they would most like to relax with” (1988: 122). In the study below I am not primarily interested in the psychological nature of reading processes, but rather in reading as an intentional act having its roots and consequences in everyday life. 3. Methodology Investigating readers’ everyday reading behavior requires access to actual readers in the most ‘natural’ circumstances, where possible bias (e.g., economical factors) is minimal. Hence I decided to investigate the clients of public libraries who access the books freely, over a convenient time span, with minimal effort (e.g., they do not have to pay for the books they choose). The main methodological problem here concerned readers’ self-idealizations of their literary attitudes. Large sociological surveys on reading seem to back up this claim (e.g., biannual surveys conducted by the Polish National Library; see Straus et al. 2004). People asked directly about their reading habits tend to overemphasize the amount of “acclaimed” readings among their choices. In order to avoid the self-idealization bias I drew on Alfred Schutz’s concept of a well-informed citizen (Schutz 1964). He argues that a well-informed citizen mediates between a man on the street (not interested in understanding more than he needs to), and the expert (who specializes in a narrow sphere of knowledge). Well-informed citizens have both practical and theoretical knowledge of a certain topic. That is why I chose librarians to be well-informed citizens in my study, to serve as informants guiding me through the everyday world of their clients – readers. The world of readers’ everyday life is complex and its exploration, I believe, requires a qualitative approach. That is why I drew on interpretative sociology, aiming at ‘understanding’ the motives behind readers everyday behavior, rather than ‘explaining’ them in a quantitative manner. Hence, the results do not aspire to be representative or statistically valid. That is the task of quantitative researchers. Yet, in this chapter I am proposing an interpretation of interviews from the perspective of phenomenological sociology, what, I reckon, sheds some light on everyday reading practices of Polish readers. In the course of the study I applied such qualitative techniques as interviews and observation. I conducted semi-structured, in-depth interviews with employees of 11 different libraries, and one participant observation in a public library. In order to differentiate the sample I decided to visit libraries in villages, small and medium towns in the Warsaw area. In large cities (Warsaw and Cracow) I chose 3 libraries from different districts to ensure the sample’s diversification.
Chapter 21. The apology of popular fiction: Everyday uses of literature in Poland
The interviews were carried on in April 2007. In selected libraries I interviewed those librarians (one per library), who had direct contact with readers. In order to learn as much as I can from librarians themselves, and to avoid framing their responses in previously set categories, I asked them open questions, concerning motivation for reading (why people read), aspects of the book choice (what readers mention, when asking for a book recommendation; do they mention different), the role of literature among other activities of the everyday life (how often their clients read and in which circumstances). The interviews were recorded and transcribed. Both transcripts and notes from the observation were interpreted in the framework of phenomenological sociology, according to discovered patterns. I am not going to discuss the study as whole. What follows is a presentation of some aspects of the study, covering the role of popular fiction in readers’ everyday lives. 4. Results: Everyday uses of literature I have already mentioned that the leisure reading is usually described as an escape from everyday problems and a mental relaxation; I reckon that Eva Maria Scherf (1990), a German cognitivist, provides better categories to capture this phenomenon. Scherf distinguishes three kinds of reception: reception as an action, activity and operation. The first two types describe the everyday reading, whereas the last one, which I leave out here, refers to a scientific activity, such as literary criticism, or academic interpretation. Reading as an action, in Scherf ’s terms, serves as an emotional stimuli or a tool to “to break down particular internal tensions (for example, by laughing)” (ibid.: 491). The other type, reading as an activity, is based on the reader’s individual biography, and its superior motive is a “desire to (re)cognize oneself and the world in appropriating a text […] the reader/viewer/listener seeks to take from the text whatever s/he can relate as directly as possible to personal practice (ibid.)”. Both those aspects are equally visible and important in the reading of popular fiction. In her study of romance readers, Radway concentrates on the emotional side of the reading process, whose main goals are relaxation and escape (1984: 88). It seems that in her study both types of reading proposed by Scherf are subject to the psychological benefits of the reader. I will compare my findings with her framework, discussing the everyday uses of literature firstly as a tool for emotional tuning, secondly, as a means of obtaining information about the world. Radway’s framework of leisure reading allows us to distinguish three aspects of the emotional impact of reading in everyday life: (1) escape from daily problems (ibid.: 93); (2) relieving tensions (ibid.: 95); (3) compensation through
Maciej Maryl
vicarious emotional nurturance and imaginative conversation with adults from a broad spectrum of social space (ibid.: 113). I conceive of these phenomena as different forms of what readers acquire through the mechanism we could call ‘losing touch with reality’, ‘being lost in a book’ (Nell 1984), or transportation (Gerrig 1993; Green & Brock 2000; Green 2004). However, I claim this mechanism is not autotelic, as a simple leisure is. Contrarily, it always has its purpose. The traveler who sets off on the journey into the fictional realm of a book, always has some basic interests (not necessarily conscious) she wants to fulfil during this trip. I am going to discuss the practical, everyday side of this psychological phenomenon in more detail, drawing on the results of my study (Maryl 2007). What follows is not an attempt to construct any typology, but rather a list of several everyday motives that guide the popular fiction reading process. The proposed perspective shows how psychological aspects of reading are employed in readers’ daily life. The motives I am going to discuss are: (a) seeking specific emotions (positive/negative), (b) emotional stimulation, (c) relaxation, (d) escape through vicarious activities. a. Processing of a literary text relies heavily on the readers’ mood and their need to alter or deepen it. This motive is described by Radway as a general aim of romance readers (1985: 90). Readers often ask librarians: “Help me find something cheerful, funny, so I could break away for a while”. Popular fiction readers often look for positive emotions, for example, this love-stories enthusiast, who said: “I had to release the tension. You know, regardless if it’s silly or not, I just had to”. Reading helps in modifying the mood evoked by everyday events. Perceiving positive events in protagonists’ lives leads to the consoling observation: ‘at least somebody (namely, the protagonist) has succeeded’. b. Reading also serves as an emotional stimulus, a sort of a natural stimulation tool. This function is very often sought by persons who work nights, such as nurses on emergency shifts, for instance. Those workers read to avoid sleeping. One librarian observed that night watchmen often come to her library, “cause they have this sort of job, that you just sit and turn stupid. And then they suddenly realize they want a book”. c. But rather it is the contrary function of reading that is the most common– literature serves as a tool of soothing emotions and facilitating falling asleep. As one of the librarians in my study put it: “I can’t sleep. To avoid lying on bed and thinking, I read”. Literature serves here as a kind of a natural sleeping pill. It is worth of mentioning that the function a text has to play determines the book choice. The text has to be simple and by no means absorbing. As one librarian admits: “nothing lulls me to sleep so well as this monotony, this stupid monotony”. Otherwise, literature serves as a stimulation.
Chapter 21. The apology of popular fiction: Everyday uses of literature in Poland
d. A different form of losing touch with reality is experiencing vicariously through reading. It occurs when a reader is deprived of the possibility of action. Entering the world of fiction provides her with similar emotions as in the case of real action. As Radway puts it, “they escape figuratively into a fairy tale” (1985: 93). We can distinguish here a temporary and a durable deprivation. A book serves as a temporary, vicarious activity when a reader travels in public transportation, for instance. She has to remain in a static position, hence concentration on events from the fictional world helps her keep the cognitive resources busy. The actual life world is torn out from the actor’s body, and it’s being removed to a different reality. If the deprivation is durable, reading substitutes an actual activity, which certain readers can no longer perform to the extent they need it. It is a vital function of reading, especially in the case of people having much spare time, such as retired persons. Literature helps to kill time, or rather to sustain the contact with the social reality people are deprived of. Reading serves as a substitute of a common social activity, providing readers with insights into biographies and motivations of other people. As I tried to show, the psychological aspects of literary response can be described in terms of a rational action undertaken in order to achieve specific goals. In other words, readers take the advantage of the properties of the reading process and make good use of them to achieve non-literary goals. The same, I would claim, goes for the other type of response, Sherf calls reading as an action. I have already mentioned the approach proposed by Radway (1984) and Nell (1988), who treat readers’ claims about the informational value of popular fiction as an excuse not to read more acclaimed books. I will claim, however, that readers, guided by their biographically determined practical interest, do seek and do find subjectively important information in popular fiction. In order to show that dimension of leisure reading, I would like to focus on those parts of my study that dealt with the relation between readers’ biographies and their choice of books. Librarians, who usually know their clients from informal conversations during their visits in libraries, observe that readers, in general, search for topics relevant to their own experience, biographies and everyday problems. Some readers, for instance, ask for books on family relationships. “It is a sort of a book”, a librarian explains, “which conveys the family values and reassures the reader”. The book serves here as a tool for attaining intersubjectivity: the reader observes a different (but similar) world and draws some conclusions from this observation. This practical interest is either directed towards the unknown content one wants to discover, or concerns the already known situations or problems that will be either illuminated from a different perspective or solved.
Maciej Maryl
In the first case the practical motive is to deepen one’s knowledge about the world by gaining information relevant to one’s own biography and hierarchy of importance. The knowledge attained in this way becomes a part of one’s experience. A good example of this function of literary reading is the reader’s search for guidance in the previously unknownsome unfamiliar situation. For instance, a reader asked one librarian about the novels on the Middle-East, because her daughter has an Arab boyfriend and the mother was worried. In the case of the inward-directed reading, the aim is to elaborate difficult situations or problems of the daily life that bother the reader in a particular moment. “If a reader suffers from an illness”, a librarian explains, “he searches for a book about it. And it doesn’t have to be a grave illness”. In a similar way teenage readers ask for books dedicated to such problems as addictions, first love or eating disorders. Such reading deepens their understanding of their own actions and attitudes. It also gives them access to both different patterns of behaviour and the reasons behind them. Although Nell claimed that practical interest in reading is rather attached to non-fiction (1984: 122), my interviews with librarians show the opposite. Readers tend to prefer to read fictional accounts rather than scientific books on aspects vital for them, be it middle-eastern culture or bulimia. This preference tells us a lot about the role of literature in acquiring new knowledge. Literature provides us with a thick, elaborated description of both depicted events and protagonists’ behavior. Hence, readers prefer to explore the world through literature, because literary reading is by no means a passive act of acquiring new information. Contrarily, it is a sort of an active dialogue of the reader’s own experience with the new content, a dialogue directed by a practical interest and facilitated by literature.
5. Conclusion As I tried to show, popular fiction reading should not be analysed exclusively as a leisurely activity. I claim that readers use this kind of literature in order to achieve some pragmatic goals of everyday life. I argued that reading in the popular fiction is guided by the biographically determined reader’s practical interest. That is to say that readers use literature as a tool for alternating and expanding their life worlds. This approach may be useful in education. Knowing the possible functions of literature should help us to pay more attention to students’ awareness of what they can find in literary texts and how they can use them on their own in order to cope with everyday life. Treating literature solely as a sociological document conveying ideology, as happened lately in Poland, is not enough, and it raises students’ negative attitudes towards literary texts.
Chapter 21. The apology of popular fiction: Everyday uses of literature in Poland
I do not claim that we should get rid of canonical readings, or abandon detailed analyses of certain works. But I do claim that in the classroom we should also pay attention to those areas of literature that are usually omitted and underestimated by literary scholars, such as popular fiction. The emphasis should not be put on the history of literature, but on the reading process, namely on how readers can benefit from works of different periods, including the canonical ones, in order to enrich their life worlds and deal better with the surrounding environment. Hence I strongly call for rehabilitation of popular fiction. It should not be overlooked by architects of reading lists. Perhaps this kind of reading does not convey the meaning memorable for the society, nor it embodies distinctive literary values. It is however still important for individual readers, since such reading plays a vital role in their experience of everyday life. Let me paraphrase Bloom, quoted earlier in this paper, all that popular fiction can bring one is one’s confrontation with one’s everyday life.
References Anderson, B. 1983. Imagined Communities: Reflections on the Origin and Spread of Nationalism. London: Verso Edtions/NLB. Bloom, H. 1994. The Western Canon. The Books and School of the Ages. New York NY: Harcourt Broad Company. Fish, S. 2004. Interpretive communities. In Literary Theory: An Anthology. J. Rivkin & M. Ryan (eds), 217–222. Malden MA: Blackwell. Giertych, R. 2007. A bill of changes in the literary canon (Polish only), http://bip.men.gov.pl/ akty_projekty/projekt_rozporzadzenia_240507.pdf. Gerrig, R.J. 1993. Experiencing Narrative Worlds. On the Psychological Activities of Reading. New Haven CT: Yale University Press. Green, M.C. & Brock, T.C. 2000. The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology 79(5): 701–721. Green, M.C. 2004. Transportation into narrative worlds: The role of prior knowledge and perceived realism. Discourse Processes 38(2): 247–266. Herder, J.G. von 2006 [1773]. Shakespeare. Selected Writings on Aesthetics. G. Moore (ed. & transl.), 291–307. Princeton NJ: Princeton University Press. Hillis-Miller, J. 2002. On literature. New York NY: Routledge. Johnson, S. 1765. Preface. The plays of William Shakespeare, London: J. & R. Tonson. Kernan, A.B. 1973. What is Literature? New Literacy History 5(1): 31–40 Maryl, M. 2007. Czym jest literatura? Potoczne pojęcie literatury i funkcje lektury w życiu codziennym czytelników (What is literature? The common concept of literature and the functions of reading in everyday life). Warsaw: Warsaw University. Nell, V. 1988. Lost in a Book. Psychology of Reading for Pleasure. New Haven CT: Yale University Press. Radway, J. 1984. Reading the Romance. Women, Patriarchy and Popular Literature. Chapel Hill NC: North Carolina University Press.
Maciej Maryl Strauss, G., Wolff, K. & Wierny, S. 2004. Książka na początku wieku. Społeczny zasięg książki w Polsce. (A Book in the beginning of the century. The social scope of a book in Poland). Warszawa: Biblioteka Narodowa. Schutz, A. 1964. The well-informed citizen. In Collected Papers. Vol. II. Studies in Social Theory, 120–134.The Hague: Martinus Nijhoff. Schutz, A. & Luckmann, T. 1973. The Structures of the Life-World (Strukturen der Lebenswelt). R.M. Zaner & H.T. Engelhardt, Jr. (transl.). Evanston IL: Northwestern University Press.
afterword
A matter of versifying Tradition, innovation, and the sonnet form in English Walter Nash “Tradition” is represented by “formalist” (“metrist”) verse; “Innovation” by “aformalist” (“free”) composition. This birthday tribute to Professor van Peer takes the formalist side, and discusses innovation and its possibilities in the context of the sonnet form. The form and its traditional variants are discussed, then ten original sonnets are presented in first draft, the opening section of a proposed sequence called Sonnets from the Wilderness. A commentary on the text, its intentions, innovations, and compositional problems, then follows.
1. Introduction … for the unquiet heart and brain A use in measured language lies.
(Tennyson, In Memoriam V)
There is a type of poetry journal, of the “vanguard” kind just now quite influential in the UK, which professes to publish only verse that is “contemporary”, “relevant” and “innovative”. If prospective contributors should ask, “contemporary with whom?”; “relevant to what?”; “innovative how?” they will usually be advised to purchase and study some back numbers of the journal to find out what sort of thing it publishes, so that they may learn to write that sort of thing, for possible publication in a contemporary, relevant, and innovative magazine. This might seem like a prescription for the begetting of sterile poems, but is intended, I believe, as a serious invitation to would-be poets to come in and learn originality by example. All serious poets (but what is “serious”?) have a wish to write original things. For some, this means “without precedent” or at least “original in intent”; which is optimistic at best, for nothing is ever quite new or wholly unprecedented. For others, originality exists as a possibility within a tradition, and derives from
Walter Nash
innovations, variations, licenses taken with patterns that have existed for maybe hundreds of years. The proponents of patterning make a poetic tribe commonly called “metrists”, or “formalists”; the counter-movement in modern poetics is towards a measured formlessness, or “free verse” . Each type, the formal and the free, has its compositional principles. The benchmark of formalism is acceptance of a controlling and coordinating system of meter, an isochronic measure, or “timing”, like beats-to-the-bar in music. In verse composition it can be expressed as feet-to-the-line, – as in the “iambic pentameter” of Shakespearean blank verse – or as syllables-to-the-line, as in the “decasyllabic couplets” of Pope’s satires. These descriptions are companion perspectives on the “movement” of English verse, which is generally reckoned to be “accentual/syllabic”. To verse-makers these terminologies are abstractions, often beyond their knowing or caring. Their immediate guide is an instinctual count, a metronome-in-the-mind, governing the realization of rhythmical patterns. The benchmark of free verse, on the other hand, is a direct commitment to the perceived phrasal rhythms of speech, without reference to a metrical “idea”. The verse so conceived is presented in lines that deliberately frustrate the operations of “scansion”. Formal and free verse have their own sound patterns, their methods of phonological focus and prominence, e.g., in directing the placing or shifting of accents, or implying changes in sonority; in these things they perhaps do not differ fundamentally, after all, the one from the other, for poetry works by many devices to common effects. Each type has its great poems; for originality in formal verse, see Owen’s Strange Meeting; for a classic in free verse, see Lawrence’s Snake. Each type has its losers; some rhymes and formal rhythms are neat and trivial beyond significance, some free verses portentous and wordy beyond endurance. The lurking treachery of formal poems is that the metric regulation puts constraints on choices of the most fitting word, the accurate phrase; the writer says something that a complex stanza form will allow him to say, which may not be what he thought he wanted to say. (though he might decide that the exercise of finding rhymewords or accentually convenient phrases has a heuristic value, pointing him to something almost as good as what he proposed). The besetting stupor of free verse is that theoretically unfettered liberty will allow the poet to say not only what he wants to say, without let or hindrance, but also what he need not say, without loss or hazard, but will go on saying, doggedly, until all compression and rhythmic power is dissipated.
. …which Robert Graves used to denounce as oxymoronic: for if it were verse, he said, it could not be free, and if free, could not be verse. T.S. Eliot’s typically Delphic view was that “no verse could be free for the man who wants to do a good job”. He considered, however, that free verse could be scanned. Graves objected that if it could be scanned, it was not free. See Eliot 1957, p.37, and Graves 1955, p.85.
A matter of versifying
Good modern poets have long been sensitive to the problem of reconciling strict formality and a-metric freedom. For a fairly recent expression, see in 1991, the (originally) formalist Thom Gunn writing to the Jon Silkin: It seems to me that we have reached an impasse right now, and there is an urgent need to reconcile meter and free verse, if that is possible, because meter is getting starved without the improvisational powers, and free verse simply turns to very dull prose without any connections to song. (Silkin 1997: 4)
This statement represents a genuine anxiety on the part of the Anglo-American Gunn, seeking to rationalize a shift in the continuum of his poetic style, from the metric precision of his English beginnings towards the stand-at-ease of a new American persona, after his emigration to the USA. He is right about the “improvisational powers” that good free verse can enhance. He is mistaken, however, in implying that formal verse lacks in itself any resort of innovation and improvisation. The history and continual renovation of lyrical forms speaks against this. There is simply no limited number of lyric patterns; poets devise them, or experiment with them continually, and every experiment in metrical form is a type of “improvisation”. Even a traditional form like the sonnet, ostensibly set in a mould too fixed for much innovation, still leaves room for experiment. 2. Ghosts from the past Scorn not the Sonnet; Critic, you have frowned, Mindless of its true honours; with this key Shakespeare unlocked his heart; the melody Of this small pipe gave ease to Petrarchs wound … (Wordsworth, Miscellaneous Sonnets, II.1)
The sonnet, as form and idea, is a traveler; it has rambled all over the globe. In origin, however, as every good European and Professor of International Hermeneutics knows, it is Italian, and is primarily associated with the name of Francesco Petrarca (or Petrarch in English), an innovator in his own time, whose love poems to an idealized mistress, Laura, are set in intricately rhymed patterns of pentameter, in fourteen hendecasyllabic lines, divisible into an octave (the first eight) and a sestet (the concluding six). The Petrarchan model has in the octave, usually the rhyme-scheme a-b-b-a a-b-b-a, and in the sestet a freer choice, e.g., of c-d-e-c-d-e, or c-d-c-c-d-c, or c-d-e-d-c-e. The fourteen lines represent in effect a complete package of “argument” (in the octave) and “response” (in the sestet). After the eighth line comes the pause and about-turn that Italians call the volta. This, purists insist, is a necessary mark of the true sonnet. But the Petrarchan sestet, in his sonnets In vita e morte di Madonna Laura, is not always a retort; it is as often the extension of a theme.
Walter Nash
Petrarch’s fourteenth century invention was carried to England in the sixteenth century. The bearer was Sir Thomas Wyatt (1503–1542), scholar, diplomat, the finest of lutenist poets, and incidentally lover of Anne Boleyn, until Henry VIII staked his claim to her. Wyatt brought back from an ambassadorial trip to Italy in 1526–7 some translations or imitations of poems from the In vita. This was one of them, a rendering of Petrarch’s sonnet XC, Pace no trovo: I find no peace, and all my war is done; I fear and hope, I burn and freeze like ice I fly above the wind, yet can I not arise And naught I have, and all the world I seize on. That looseth nor locketh holdeth me in prison, And holdeth me not, yet can I scape nowise Nor letteth me live or die at my devise And yet of death it giveth no occasion. Without eyes I see, and without tongue I plain; I desire to perish, and yet I ask health I love another, and yet I hate myself; I feed me in sorrow, and laugh in all my pain. Likewise displeaseth me both death and life, And my delight is causer of this strife [“That” in line 5 = “that which”, i.e., “what, whatever”]
As Pace no trovo is a love-plaint to Madonna Laura, so this complains to Ms Boleyn. It is by no means a slavish imitation, and could never serve as a crib. It has some moments of verbal ingenuity: for example, “all the world I seize on” in line 4, glimpses at “seizin”, “take seizin”, an Anglo-French legal term (still in use), meaning “take freehold possession of ”. The meter, however, presents problems to all strict counters of beats and syllables. Critics have damned or apologized for Wyatt´s metrical procedures in these and other poems (e.g., the great lyrical plaint, “They flee from me that sometime did me seek”, another Ann Boleyn poem). They speak of “awkwardness” and “tentativeness”, or of the sonnets as “mere exercises … roughly jotted down in whatever broken rhythms came readiest to hand” (see Hollander & Kermode, 1973, p.116; and Chambers 1933, p.122). In his Petrarchan sonnet-versions, Wyatt was attempting to adapt eleven-syllable lines to what instinct told him was a decasyllabic norm for English. He wrote, furthermore, as an instrumentalist, with a concept of performance that admitted of changes of tempo, “slurs” and “rests”; above all, what he heard as he wrote was the magical metronome-in-the-mind, spoken of above, the unarithmetical measure that adjudges to itself the rightness of a reading or a speaking. His translation of Pace no trovo requires no apology on technical grounds, it stands up as an English poem. It is particularly remarkable in one respect, if we compare the
A matter of versifying
rhyme schemes in the two poems. Pace no trovo has this: a-b-a-b a-b-a-b/c-d-e c-d-e – the strict Petrarchan norm. But I have no peace has this: a-b-b-a a-b-b-a/ c-d-d c-e-e, with a sestet ending in a couplet, so emphatically that we might call it a “clinching sestet”. None of the In vita sonnets clinches a poem in this way. And because it has no obvious turnabout, no volta, the scheme of I have no peace might be read as a-b-b-a/a-b-b-a/c-d-d-c/e-e . It is, indeed, a Petrarchan sonnet “ghosting” the form-to-come of the Shakespearean. This was the scheme that Wyatt devised, and that Wyatt’s friend and poetic imitator, Henry Howard, Earl of Surrey used in his own, graceful, strictly metered sonnets. The “clinching sestet” was the prescription for numerous sonnets, indeed for the majority of those written in the sixteenth and early seventeenth centuries. The greatest of Elizabethan sonneteers, Sir Philip Sidney (1554–1586) used it for many of the poems in his great sonnet cycle, Astrophel and Stella. But he was not a man to follow trends tamely. The first poem in his Petrarchan sequence is written in alexandrines, French-flavored dodecasyllables, not easy to manage in English. Sidney manages them expertly, all the way through to a clinching sestet: Loving in truth and fain in verse my love to show That she, dear she, might take some pleasure of my pain. Pleasure might cause her read, reading might make her know, Knowledge might pity win, and pity grace obtain. I sought fit words to paint the blackest face of woe: Studying inventions fine, her wits to entertain, Oft turning other’s leaves, to see if thence would flow Some fresh and fruitful showers upon my sunburned brain. But words came halting forth, wanting Invention’s stay, Invention, study’s child, fled stepdame Nature’s blows, And others´ feet still seemed but strangers in my way. Thus, great with child to speak, and helpless in my throes, Biting my truant pen, beating myself for spite, ”Fool!” said my Muse to me, “look in thy heart and write”.
This is an extraordinary declaration of independence, by a man resolved not to read back numbers of anyone’s pamphlet and try to do likewise, but to think and feel for himself. Sidney knew in all their complexity the problems which poets still have: of keeping or quitting a tradition, of using or ignoring models, of courting the artifice of creativeness, of responding to the rough prompts of one’s own temperament (“stepdame Nature’s blows”). He invokes the conventional image of the laboring poet (“great with child to speak, and helpless in my throes”), impatient and self-accusing in the discovery of his failures (“biting my truant pen, beating myself for spite”). Sidney, for sure, knew Petrarch, knew the tradition; but more, knew the problems of escaping from it, or not being enslaved by it. Here in this
Walter Nash
first sonnet of a sequence, the man is declaring for his own Muse, i.e., his own way of doing things. And it is to this, not to his emblematic Stella (her emblem the star – as Petrarch’s Laura flourished in the emblem of the laurel tree) that the first of the 108 sonnets by Astrophel – “star-lover” – is addressed. What follows is an accomplished sequence, reading the implications of a romantic fiction which is the mask, or semblance, of a restless mind. Tucker Brooke´s comment on it is witty and just: “It is as Hazlitt said of Romeo and Juliet, the story of Hamlet in love” (in Baugh 1950: 479).
3. Novelty and excitement The next two decades, from 1590 to 1610, were to be the heyday of the Elizabethan sonnet. Merely to list the sonnet sequences written at this time suggests that now the sonnet is contemporary, innovative, and relevant, an art form for people in the know: Samuel Daniel’s Delia, 1592; Michael Drayton’s Idea’s Mirror; Amours en quaterzaines, 1594; Edmund Spenser’s Amoretti, 1595; and in a genre apart from these, John Donne’s Holy Sonnets, circa 1609. In form, all follow the rhyme-pattern introduced by Wyatt and Surrey and generally used by Sidney, the “clinching sestet” that rolls up to an end-rhyme. Spenser’s sonnet has, in addition, a scheme of interlocking rhymes, through to the closing couplet: a-b-a-b b-c-b-c c-d-c-d e-e. In “tone of voice”, or “stance” they differ greatly, each sonneteer his own persona in his own relationship to the plausible fiction and inner depth of a theme. The reader of Spenser will discover a courtly refinement in, say, One day I wrote her name upon the strand [Amoretti, lxxv]; of Drayton, a robust street-style in Since there’s no help, come let us kiss and part [Idea, lvi}; and in John Donne, Dean of St. Pauls, a pious fervor, like a Welsh preacher’s hwyl, voluble and passionate in, for example Batter my heart, three-personed God [Holy Sonnets, x]. In a few years, after Sidney’s exemplary prompting, the sonnet has made serious progress. It has also, by this point, progressed as far as the complicated Mr. William Shakespeare, whose Sonnets, collected and published in 1609, were presumably composed over a period of years. They are all written to a personal rhyme scheme, possibly derived from the “clinching sestet” model, but idiosyncratic in its distribution of the poetic theme across three quatrains and a concluding couplet, e.g., a-b-a-b/c-d-c-d/e-f-e-f/g-g. The dating and compositional order of the poems is a problem. Yet more enigmatic is the theme of this long recital of 154 poems. They are not fixed on a conveniently fictitious ladylove. They speak of male friendship, of homosexual relationship, of adulterous relationship with a promiscuous “dark lady”, of admiration, jealousy and suspicion of a poetic rival, of pleasure, of shame, of illness, of depression – of such things as are the common stuff of truelife
A matter of versifying
theatre. “With this key, Shakespeare unlocked his heart”, Wordsworth said; to which Browning shrewdly retorted “Did Shakespeare? If so, the less Shakespeare he!” His mature sonnets are psychological drama of a dark, uneasy sort, the work of a playwright. As early as 1595 he was making sonnets for dramatic use, in the tragedy of Romeo and Juliet, where the first and second Acts have Prologues in the form of sonnets, and where a sonnet is actually involved in the dialogue of the play leading to the tender, sportive dialogue of the lovers’ meeting. In a paper on aesthetic criteria in the formation of a literary canon, Professor van Peer has drawn attention to this passage from the first act (I, v, 92–110) (van Peer 1996: 106): Romeo:
If I profane with my unworthiest hand This holy shrine, the gentle sin is this: My lips, two blushing pilgrims, ready stand To smooth that rough touch with a tender kiss.
Juliet:
Good pilgrim, you do wrong your hand too much, Which mannerly devotion shows in this; For saints have hands that pilgrims´ hands do touch, And palm to palm is holy palmers´ kiss.
Romeo: Have not saints lips, and holy palmers too? Juliet:
Aye, pilgrim, lips that they must use in prayer.
Romeo: O then, dear saint, let lips do what hands do: They pray: grant thou, lest faith turn to despair. Juliet: Saints do not move, though grant for prayer’s sake. Romeo: Then move not, while my prayer’s effect I take.
(He kisses her)
The young lovers, playfully approaching their first kiss, are made to speak a sonnet, in the form a-b-a-b/c-d-c-d/e-f-e-f/g-g. Romeo speaks the first quatrain, Juliet the second, they share the third, and they have alternate lines in the clinching couplet. We have to suppose that some, at least, of the Elizabethan theatre audience would recognize what they were hearing; they would know a sonnet when they heard one, and they would be sensitive to its charm, its innocence, and its foreboding irony in this context. 4. Beyond the golden age Knowing the heart of man is set to be The centre of his world, about the which These revolutions of disturbances Still roll; (Samuel Daniel, Epistle to the Lady Margaret, Countess of Cumberland)
Walter Nash
This Tudor-Jacobean time was the golden age of the English sonnet, when, it might seem, everything had been done with the form, and it had been put to doing everything it could do. Of course, the possibilities of the sonnet, as variable discourse, were far from exhausted. The centuries that followed brought interludes of sonnetry, periods of cultivation or recession in response to the mood and ideology of the time. (A notable dry interlude is the eighteenth century.) We have had our sessions of great sonneteers: in the seventeenth century, Milton, talking stern morals and republican zeal; among the Romantics, Wordsworth, talking of humble rivers and huge patriotism; among the Victorians Elizabeth Barrett Browning, role-playing the beloved adoring the lover, and incidentally putting in a word for feminism; in the twentieth century W.H. Auden with an unrivalled Petrarchan cycle of 27 poems, In Time of War, naming nor lover nor mistress, but rehearsing, with wry Stoic eloquence, the Fall of Man. The sonnet lives on, the poems continue to come, whatever their theme might be. Their properties are those our English folk-saying recommends for bridal wear on a wedding-day: “something old, something new, something borrowed”, and one more, “something blue” – for the bride, maybe, something concealed, for the poet, something to be found, a touch of originality, which, as noted at the onset of these musings, all serious and even jesting poets covet. 5. A tribute Enough, if something from our hands have power To live, and act, and serve the future hour … (Wordsworth, Sonnets on the River Duddon, xxxiv)
What remains to be said is personal; a few words addressed to the distinguished Herr Professor Dr. van Peer, my learned acquaintance, my deeply valued friend, Willie, onlie begetter of so many tributes, on the occasion of his sixtieth birthday. I would like, sir, to conclude my own tribute by giving you the first sight of the first draft of a set of sonnets called “Sonnets in the Wilderness”. This set, a part of a planned larger sequence, is, you may say, an experiment in controlled innovation. My sonnets at times respect the form inherited from Petrarch & Co., and at times drastically modify it, bringing into question the whole idea of what a sonnet is, or could be. I have treated it in some places as a lyric fourteener, with rhymes, in one place as an unrhymed quatorzeine; I have in some places argued by octave and sestet, in others ignored the distinction; I have used disyllabic rhyme, internal rhyme and pararhyme (half-rhyme, slant rhyme). I have not yet started to use designed schemes of alliteration, but that may well come. In short, Willie, I am pushing the envelope, as the current jargon goes, and waiting to see how it crumples. “Sonnets
A matter of versifying
from the Wilderness” has two concerns: (a) the isolation-in-mind, the exilic state of a bookish man, and (b) a reaching, half-resisted, desperately felt, for religious commitment. I do not mean to suggest that a man as outgoing and interactive (in several languages) as you, must ever feel isolated; and as for God, you are far too young, at the age of sixty, to be thinking about any such thing (I paraphrase Mistress Quickly). I hope, though, that you will recognize my voice and gesture in these poems, and take some pleasure in doing so. Have a happy, blessed, birthday, Willie. Tot siens. Bill. SONNETS FROM THE WILDERNESS (a draft: for Willie van Peer) I Not sleeping well; the summer nights are bad, bed has no comfort for the bones, it seems; but worse, these fretful unforgiving dreams fraught with misgivings; I am old and sad. At seven I’m up; let in the Siamese bawling his need to be adored and fed; make tea; consider going back to bed then can’t be bothered. Stirred by the morning breeze, fronds of Canary palm arrest the light breaking out over the hill behind. I bless the bluetits perching on my breakfast tray. Serene tomorrow issues from last night, provisional heaven and earth begin their day. Hail, Holy Light. Good morning, Wilderness [“Hail, Holy Light” – see Milton, Paradise Lost, Bk. III, line 1.; “Canary palm”, Phoenix canariensis, a native to the islands; I have two in my garden. This is a quatorzaine, a “fourteener”, with sonnet-rhymes but no octave-sestet pattern; a sequence of remarks, rather] II Not like those ragged cliffs, that tangled steep of stone and shale and poorgrass where a saint might scramble in an ecstasy, and faint for want of bread, or mazed for lack of sleep, blunder his destined way to paradise, a holy man gone crazy by design – in no such wilderness I wander; mine’s a strange glass-flowering tract behind the eyes, a garden of abstractions where the fruit
Walter Nash
is artificial, taste a paradigm of appetising structures, nothing real; no ground footprinted by the sturdy brute, a blank half-acre of the mere ideal, a loneliness I till, to pass the time. [“those ragged cliffs”, the enormous Acantilado de los Gigantes, presiding loftily over the end of the street where I live: “poorgrass”, meaning “scrub”, a word I invented to ease the metre; Poorgrass is actually the name of a rustic character in Hardy’s The Mayor of Casterbridge: “till”, as in “cultivate”. There is no volta at line 8–9. The sense runs unbroken in two long sentences, from line 1–7, and from 7–14] III We bookish men, confiding in the mind, soon lose the confidence of what we mean; thought-galleries are darker than a mine and every pitfall makes a breach to mend; in daily rigour each one works his stint at thesis, doctorate, edition, tract, into this pious torment they are tricked – hard labour, hardly better than a stunt to have a fellowship, a high degree, a reputation among “them that know”, a senseless sense of being an elite; when recognition comes, it comes too late, too late the honoured scarlet, nothing’s new in no-man’s-land; the common wear is grey. [On intellectual isolation – and vanity: the innovation here is the use of halfrhyme throughout, echoing the mood of uncertainty, frustration, incompleteness; “stint at thesis, &c”, cp. the miner’s allotted daily stint at the coalface; “stunt”, as in “spectacular trick”; “them that know”, echoing Dante on Aristotle, Il Maestro di color che sanno, “The Master of them that know”, Inferno, iv, 131] IV But when the sea comes hero-vaulting over the harbour wall hurling up crests and canopies of spray, or when it lies, languid and tinsel-threaded in the bay, something might be attempted, after all, something not done with “do”, when simply “be” springs a surprise.
A matter of versifying
The world I see glimmering at me with its droll blue gaze begging a poem, coming to me for food, trances me sometimes with its artless ways; then suddenly wilderness softens into solitude. [A totally innovative “fourteener” with octave and sestet in lyric form, not sonnetform; “arguing”, however, like a conventional sonnet; “the sea”, “the harbour”, “the bay”, like “those ragged cliffs” in sonnet II, features I see every day, markers of my “world”, not my “wilderness”] V The accidence of nature, the small cues clenching creation’s syntax show me how a pattern is evolving even now. “The fool hath said in his heart, There is no God.” God’s poem is in the making, and yet – whose design is here, whose eye construes the text? Reader, read on, spell out what happens next; and still blind reason says, there is no God. No God exists, because no God can be held in the labouring mind’s prolixity, juggling incessantly with hope and fear, finding nor rest nor reason to rejoice in any certainty, until a voice says gently to the heart, “But I am here”. [“accidence”, as in grammar, but punning casually on “accidents”: the rhymescheme in the octave is unusual, a-b-b-c/a-d-d-c; I mark the volta here, the transition to the sestet, where “But I am here” answers the octave’s repeated “There is no God”] VI To church I go, and break out in a rage; “they manacle the sinner with their cant, they fill the hungry with their naked want, fears they inspire, in order to assuage, doubts they provoke, in order to remove faults they impute, in order to forgive, hell they imply, in order to reprieve, then speak incomprehensibly of love.”
Walter Nash
“And what is wilderness, compared to this? here is a comely falsehood, a sweet show a Punch and Judy pitched at heaven’s gate.” But as I speak, with cackling emphasis, a voice in me says “wait, and you will know; come in; be still; prepare for nothing; wait.” [A turning point in the set, this marks abeginning of a religious experience: “break out in a rage” – cp “break out in a sweat”, “break out in a rash”; also “break out from”, “escape from (punitive) restraint”. In mundane fact, I had been to church and came home in a temper. “Punch and Judy”, a marionette show for children, Kasperleteater, my mockery of Anglo-Catholic ritual. The last three lines remind me of George Herbert, but I cannot cite the parallel] VII Waiting is what humans do; we are wired for it. Flowers are not patient; tolerant only of the sun’s periodic attentions, they open and shut without giving it a thought; and my cat, Frank, (the Siamese) can’t wait, won’t wait, he wants his cuddle or chickenbits NOW, he cancels time when he blinks his eyes; as for the bluetits, time is the nanosecond between migrations. But we are different; we know how to wait for justice, or revenge, or love, or change. Patience we have – of all the things we do humans can “do waiting” – it’s our talent, also our tragedy; hung on the blind event, all the bells toll for us – abide, abide. [Totally innovative – a sonnet without rhyme, and in the language and rhythms of prose or conversation; a freeverse fourteener, but with a differentiated octave and sestet. Frank the Siamese, and the bluetits, are from sonnet I. “All the bells toll” in the last line is an accidental reminiscence of Donne’s “for whom the bell tolls”] VIII It’s hard to abide myself; I am so weak I cannot sleep for waking, and my bones practise the craft of aching; when I speak I squeak, or munch in demi-semitones. Hearing has quit my service, and my sight betrays the light, the vision’s blurred and small, and none too tall to wage a losing fight.
A matter of versifying
I try to stand, knowing I have to fall. And always wearying for a gift of grace to help me leave my foolishness behind, the mirror knows the worry in my face. I hear the gentle murmur in my mind – “for all the days amiss, the doing marred, accept your life – if soul remain unscarred.” [In the octave, a frolic of internal rhymes, playful in the misery of old age. This is the only sonnet of the set to use the “clinching sestet”] IX A little, a little at a time, take it steady now, a little to walk on the flat, a little to climb as you are ready now, walk slowly, but walk with deliberate tread towards the next turn, you may never see happenings that lie ahead, but they concern you. Don’t doubt it, never begrudge this march towards mystery swaddled in sackcloth at the far end of the story; think, how tremendously little by little, we trudge the undiscovered path, the gangway to glory. [As in nr. V, this replaces the conventional sonnet form with the idea of the lyric; octave and seset are like the introduction and refrain of a theatrical song – but they remain octave and sestet; “gangway” – as “walkway”, not as “gangplank” (!)] X Walk out into the world of heaven’s making; paths that the prophets trod – the holy men, the very elect of God – they are still yours for the taking. Make no provision for the fret of living think that necessities are placed wherever is need in the unrelenting waste. God is forever giving and forgiving. Even the wilderness you yourself created, a Fatherly omniscience invades.
Walter Nash
In all the cities of bleak memory garrisons of the Ghost are concentrated, Christ saunters on the fractured esplanades, the paper shops sell maps of Galilee. [An attempt at serenity, “relaxation”; lines of irregular length; rhymes balancing feminine and masculine, e.g., “making-trod-God-taking”, “living-placed-wasteforgiving”. The sestet invokes the Trinity; “cities of bleak memory”, episodes of life that sees empty and meaningless in recollections (or dreams)]
References Baugh, A.C. (ed.). 1950. A Literary History of England. London: Routledge. Chambers, E.K. 1933. Sir Thomas Wyatt and some Collected Studies. London: Sidgwick & Jackson. Eliot, T.S. 1942[1957]. The music of poetry. In On Poetry and Poets, 26–38. London: Faber. Graves, R. 1955. The Crowning Privilege: The Clark Lectures 1954–55. London: Cassell & Co. Hollander, J. and Kermode, F. (eds). 1973. The Oxford Anthology of English Literature. Vol. II: The Literature of Renaissance England. Oxford: Oxford University Press. Silkin, J. 1997. The Life of Metrical and Free Verse in Twentieth Century Poetry. New York: Palgrave Macmillan. Van Peer, W. 1996. Canon formation: Ideology or aesthetic quality? The British Journal of Aesthetics 36(2): 97–108.
About the contributors
Anna Chesnokova is Associate Professor at the Translation Department of Kyiv National Linguistic University, Ukraine, where she was Head of the English Philology Chair for three years. She is moderator of the Ukrainian chapter of edis (Emily Dickinson International Society). She has an M.A. in English and French Philology from Kyiv National Linguistic University and a Ph.D. in Comparative Literature from Kyiv National Taras Shevchenko University. She has published on cognitive poetics, stylistics and literary awareness and has been one of the coordinators of the international redes Project since 2003. Her specific research interest is Emily Dickinson’s poetry. Recently, she has started her post-doc on poetic awareness. Arthur C. Graesser is Full Professor in the Department of Psychology and coDirector of the Institute for Intelligent Systems at the University of Memphis. He has worked in several areas of cognitive science, artificial intelligence, and discourse processing, including text comprehension, inference generation, conversation, question asking and answering, tutoring, and advanced learning environments. He is currently Associate Editor of the journal Discourse Processes, was Senior Editor of the 2003 Handbook of Discourse Processes, and the next Editor of Journal of Educational Psychology. He has designed and tested cutting edge software in learning, language, and discourse technologies, including AutoTutor and Coh-Metrix. Bill Louw is Chair of the English Department of the University of Zimbabwe. His main interest lies in the area of stylistics. He effectively started corpus stylistics as a discipline by publishing the first article in this area in 1987. His work with John Sinclair in the area of semantic prosodies has led to his current interest in the interface between collocation, truth and the philosophy of language. He believes that collocation is nascent instrumentation for language and that its use as instrumentation will either automate or falsify linguistic theories and schools of thought. Bin Zhang received his undergraduate degree in Computer Science from Beijing University of Aeronautics and Astronautics and his M.Sc. degree in Mechanical Engineering from The University of Memphis. He is currently pursuing an M.Sc. degree in Computer Science at the University of Memphis. His current research interests are in computational linguistics and language generation.
Directions in Empirical Literary Studies
Brent Morgan is a doctoral student in the Institute for Intelligent Systems at the University of Memphis. His general interests are in cognitive science, with specific research in the areas of intelligent tutoring systems, animated conversational agents, and learning environments with automated natural language comprehension. He has conducted research with Coh-Metrix, AutoTutor, aries, and other systems with natural language technologies. Colin Martindale received his Ph.D. in Psychology from Harvard University in 1970 and a Doctorate Honoris Causa from the Université Catholique de Louvain in 1988. He is currently Honorary Professor of Psychology and Art at the Perm State Institute of Arts and Culture (Russia) and Professor Emeritus of Psychology at the University of Maine. He has served as President of the International Association of Empirical Aesthetics and of the Society for the Psychology of Aesthetics, Creativity and the Arts. He has received numerous awards for his over 200 books, articles, and chapters on creativity and quantitative studies of art and literature. David L. Hoover is Professor of English at New York University, where he teaches stylistics, English language, and Humanities computing. His publications in stylistics and Humanities computing include three books – A New Theory of Old English Meter, Language and Style in The Inheritors, and Stylistics: Prospect and Retrospect – and numerous articles on authorship attribution and corpus and computational stylistics. Active in Humanities computing for more than twenty-five years, he was elected Vice President of the Association for Computers and the Humanities in 2006. He is currently co-editing an electronic scholarly edition of Chaucer’s Clerk’s Tale, and writing a book on chronological changes in authorial style using corpus and computational methods. David S. Miall is Professor of English at the University of Alberta in Canada. Previous publications include, as Editor, Humanities and the Computer: New Directions (1990), Romanticism: The CD-ROM (1997), and a monograph, Literary Reading: Empirical and Theoretical Studies (2006). He specializes in literature of the British Romantic period, and the empirical study of literary reading – a field in which he has collaborated with Don Kuiken (Department of Psychology) since 1990. He teaches courses in Romantic literature, Gothic fiction, literary computing, and empirical and historical studies of literary reading. Donald C. Freeman is Emeritus Professor of English at the University of Southern California and co-Director of the Myrifield Institute for Cognition and the Arts in Heath, Massachusetts, USA. He has published widely in stylistics, metrics, and cognitive metaphor, and was Editor-in-Chief of pala Papers, a series of volumes published for the Poetics and Linguistics Association. In 1971 Freeman founded
About the contributors
the Department of Linguistics at the University of Massachusetts, Amherst, where he remains an Adjunct Professor. Freeman edited two volumes of essays in stylistics: Linguistics and Literary Style (1970) and Essays in Modern Stylistics (1981). Don Kuiken holds a Ph.D. and is a psychologist at the University of Alberta in Edmonton, Alberta, Canada, specializing in the study of dreams, aesthetics, and phenomenology. He has published a number of journal articles and book chapters concerning self-transformation through significant dreams, aesthetic experience, and intensive self-reflection. His current research addresses shifts in the sense of self that can sometimes occur through literary reading and as a result of impactful dreams, especially following loss or trauma. Geoff Hall is Head of Applied Linguistics at Swansea University, Wales, U.K. and Assistant Editor for the journal Language and Literature. Research interests are language in education and literary stylistics. Recent publications include Literature in Language Education (Basingstoke: Palgrave 2005), ‘Literature as a social practice’ (in The Art of English. Literary Creativity, S. Goodman and K. O’ Halloran (eds). Basingstoke: Palgrave 2006), and ‘Stylistics in second language contexts’ (in G. Watson and S. Zyngier eds. Literature and Stylistics, Basingstoke: Palgrave 2006). Ildikó Somogyváry graduated from the University of Pécs where she is currently a part-time Ph.D. student. Her research interests include clinical psychology and narrative psychology with particular emphasis on empathic processes. She is a part-time Assistant Professor in Psychology at the Eötvös College, Baja. Jan Auracher is a lecturer at the University of Munich, where he gives introductory courses in empirical methodology and statistical analysis in intercultural studies. He studied German as a Foreign Language, computer science, and neurobiology at the University of Munich, Germany, and during as well as after his studies he has been a teacher of German in Japan, Korea, and Canada. His doctoral work was on psychophysiological measurements of reactions to suspense. Currently he is involved in validating computer based text analysis through the observation of psychological and bio-psychological reader response. In addition, he is engaged in a comparative study of German and Japanese information processing, regarding reading as well as writing. Jan Prasil is a student of political science, philosophy and Germanic studies at the University of Munich. He holds his B.A. in political science from Charles University, Prague. His work examined the post-modern concepts of power in international relations. His main academic interests are post-modern readings of world
Directions in Empirical Literary Studies
politics and intercultural hermeneutics. He has been a member of the Association for International Affairs in Prague since 2006. János László is Professor of Social Psychology at the Institute for Psychology of the Hungarian Academy of Sciences, Budapest and the University of Pécs, where he is currently Head of the Institute of Psychology and of the Doctoral School in Psychology. He has been Visiting Professor at several North-American and European universities. His major research interest is psychology of narratives. His books include Cognition and Representation in Literature (Akadémia: Budapest) and The Science of Stories (Routledge: London). Jèmeljan Hakemulder has a background in literary theory and comparative literature. For his Ph.D. he specialized in the Psychology of Literature, focusing on the effects of reading literary texts on self-concept and social perception (The Moral Laboratory. Amsterdam: Benjamins, 2000, supervised by Fokkema and Van Peer). He now teaches psychology of media and the arts at Utrecht University, and trains students in the Humanities in research methods of the social sciences. In his research he looks at reception of film (e.g., response to foregrounding, and adaptation), but also still takes an interest in the effects of literary stories. Keith Oatley is Professor Emeritus in the Department of Human Development and Applied Psychology, University of Toronto. He is a Fellow of the Royal Society of Canada, and a Fellow of the British Psychological Society. His research is on the psychology of emotions and the psychology of imaginative literature. He is the author or co-author of some 150 journal articles and chapters, and six books on psychology. He is also author of two novels, one of which, The case of Emily V., won the 1994 Commonwealth Writers Prize for Best First Novel. Maciej Maryl is an assistant lecturer at the Institute for Literary Research of the Polish Academy of Sciences (Warsaw) and a Ph.D. student at the Graduate School for Social Research (Warsaw). He graduated with distinction from Warsaw University in sociology (2007) and literature (2006). His Ph.D. project is aimed at describing the role of literature in the context of new media. He works for a scientific journal Teksty Drugie. He also translates articles of such authors as Mieke Bal or Claire Cavanagh. His research interests include literature to everyday life and qualitative approaches to empirical studies of literature. Maria Dudusova is a student of Germanic studies, English linguistics and literary studies at the University of Munich, Germany. She has recently finished her B.A. Her academic interests are modern American literature and linguistic pragmatics.
About the contributors
Marisa Bortolussi is Full Professor in the Department of Modern Languages and Cultural Studies at the University of Alberta, Canada. She has an M.A. in comparative literature from Carleton University, and a Ph.D. in Spanish from Laval University. Her areas of specialization include literary theory, modern narrative, Hispanic literature, and comparative literature. For the past thirteen years she has been engaged in a collaborative, interdisciplinary research program with Dr. Peter Dixon of the Psychology Department at the University of Alberta. In this research they apply the methods of cognitive psychology to the study of literary response. Together they published Psychonarratology: Foundations for the Empirical Study of Literary Response (Cambridge University Press, 2003). Maja Djikic is a Post-Doctoral Fellow in the Marcel Desautels Centre for Integrative Thinking, Rotman School of Management, University of Toronto, having recently completed a period of work as a Post-Doctoral Fellow in the Psychology Department, Harvard University. Her research is on the effects of art on the transformation of personality. She has published in Creativity Research Journal, Personality and Individual Differences, Journal of Consciousness Studies, Journal for the Theory of Social Behaviour, and elsewhere. Mariya Sergeyeva has an M.A. in English and German Philology from Kyiv National Linguistic University. During student years, she was the Head of unesco Student’s Club and one of the co-founders of redes-Ukraine in 2003. Since then she has helped to organize 4 annual redes conferences in Kyiv and herself participated in 8 international conferences in Germany, Spain, France, Morocco, Poland, promoting redes ideas and presenting her research. In 2007, she contributed to Acting and Connecting: Cultural Approaches to Language and Literature, edited by Sonia Zyngier, Anna Chesnokova, and Vander Viana. Max M. Louwerse is Associate Professor in the Department of Psychology and the Institute for Intelligent Systems at the University of Memphis. He received M.A. degrees in language and literature from the University of Utrecht and a Ph.D. degree in linguistics from the University of Edinburgh. He has published over 70 articles in a variety of journals and proceedings, co-edited (with Willie van Peer) an interdisciplinary volume on thematics and received awards for his teaching and research. His interests cover a wide range of topics in interdisciplinary research related to computational psycholinguistics, including text and discourse comprehension and production and multimodal communication. Michael Kimmel is a researcher based at the University of Vienna, Austria. He holds an M.A. (1995) and a Ph.D. (2002) from Vienna University. His interests
Directions in Empirical Literary Studies
comprehend cognitive linguistic methods of metaphor and image schema analysis, socio-cultural embodiment, cognitive narratology, as well as qualitative and mixed methods. In the recent years he has conducted research on metaphor in political discourse with qualitative software tools. Complementing earlier psycholinguistic work on sensorimotor resonance in reading, he is currently working with text-linguistic methods on imagery in literary cognition. From early 2008–10 he will conduct cognitive-ethnographic fieldwork on embodied imagery in tango, in a project that combines linguistic and phenomenological methods with motion analysis to explore dance class interactions and body learning. In the past, Kimmel has been member of the Austrian Academy of Sciences, the Institute of Advanced Studies, Vienna and the University of Economics, Vienna, and he has worked as a freelance researcher in Vienna, Budapest, and Berlin. Milena Mendes holds an M.A. in applied linguistics. Currently, she is working towards her Ph.D. in language studies at the Catholic University of Rio de Janeiro. She has been a member of the redes group since its beginning, which has contributed to her academic background. Her interests include research on teaching English as a foreign language, corpus linguistics and the influence of technology on educational settings. Natália Silveira holds a B.A. in English and Portuguese language and literature from the Federal University of Rio de Janeiro. She has been a member of the redes Research Group since 2004 and published articles in the fields of text linguistics and computational stylistics. Nick Benesh is a postgraduate student at the University of Memphis in the Psychology Department and the Institute for Intelligent Systems. He is currently in the Ph.D. program. His research interests primarily focus around vision and how it is used in various multimodal settings. More recently his interests have included spatial visual processes and visual plasticity in transferring from video game environments to real world environments. Olivia Fialho is a Ph.D. student in the Comparative Literature Program at the University of Alberta, Canada. She has completed an M.A. with distinction in applied linguistics at the Federal University of Rio de Janeiro, Brazil. She is a member of igel and of redes. Her most recent publications include contributions to Literature and Stylistics for Language Learners (2007), edited by Greg Watson and Sonia Zyngier, and to the special issue on foregrounding in Language and Literature (2007). Her Ph.D. project is an interdisciplinary study of self-modification in literary response. Her research interests include the empirical study of literary reception, modernist fiction, and literary awareness.
About the contributors
Paul Sopčák holds an M.A. in English literature from the University of Munich, Germany, and is currently pursuing a doctoral degree (abd) in comparative literature at the University of Alberta, Canada. He is a Killam Scholar whose academic interests include the empirical study of literature, early modernist English literature, and Latin American literature. He has published on readers’ responses to stylistic features in drafts to James Joyce’s Ulysses, on responses to depictions of violence in Latin American literary texts, and on intertextuality in Bruce Chatwin’s Songlines. Peter Dixon earned his Ph.D. from Carnegie Mellon University in 1979. After a two-year post-doctoral position at Bell Laboratories, he moved to the University of Alberta where he has been since. He has published broadly in cognitive psychology, including research in areas such as word recognition, discourse processing, working memory, visual attention, visual perception, motor control, and statistical methods. With Marisa Bortolussi, he has co-authored Psychonarratology: Foundations for the Empirical Study of Literary Response. Previously, he has served as Editor of the Canadian Journal of Experimental Psychology and as Associate Editor of Memory & Cognition. Raymond A. Mar is an Assistant Professor at York University in the Department of Psychology. His work has appeared in various journals including Neuropsychologia, Journal of Research in Personality, Social Cognitive and Affective Neuroscience, Perspectives on Psychological Science, and in the book Empathy and Fairness (Wiley & Sons). The central topic of his research is the simulation of social experience afforded by fictional narratives such as literature, cinema, and theatre. Sonia Zyngier is Associate Professor of English Language and Literature at the Federal University of Rio de Janeiro, Brazil. She holds a Ph.D. in Applied Linguistics (University of Birmingham). In her thesis, she developed the concept of literary awareness, which has since been validated in different countries, and published widely. Her latest publications include Muses and Measures: Empirical Research Methods for the Humanities (2007), in collaboration with Willie van Peer and Jèmeljan Hakemulder, Acting and Connecting: Cultural Approaches to Language and Literature (2007), edited with Anna Chesnokova and Vander Viana, and Literature and Stylistics for Language Learners (2006), edited with Greg Watson. She has also written on pedagogical stylistics for the Elsevier Encyclopedia of Language and Linguistics (2006). Uri Margolin is Professor Emeritus of Comparative Literature at the University of Alberta, Edmonton, Canada. His main areas of research and publication are literary theory and methodology, narrative studies (including possible worlds
Directions in Empirical Literary Studies
semantics and cognitive approaches) and Slavic formalism, structuralism and semiotics. Publications to date include close to 70 essays in collective volumes and professional journals. His association and friendship with Willie van Peer go back many years to van Peer’s first visit to Edmonton. Vander Viana holds a B.A. in English language and literature from the State University of Rio de Janeiro and is currently doing postgraduate work at the Catholic University of Rio de Janeiro. Since 2003 he has been a member of the redes Research Group. He has co-edited five volumes of collected works and published articles on corpus linguistics, English language and distance learning. Acting and Connecting: Cultural Approaches to Language and Literature (2007), edited in collaboration with Sonia Zyngier and Anna Chesnokova, is among his latest publications. Vladimir Yepishev graduated from Kyiv National Linguistic University with Specialist Degree in translation and interpretation (English and German). In 2006, he joined the Ukrainian redes Group, and presented a paper on Tolkien’s novels at the redes Conference in 2007. He is currently working on a new project on sound symbolism. He has published an article entitled “Dwarfs and Hobbits in Ukrainian Translations.” Walter Nash is Emeritus Professor of English Language, University of Nottingham. Since his retirement in 1991 he has lived for long periods in Tenerife. His main occupation of late has been writing verse, e.g., Of Time and Small Islands (a book about landscapes, histories, myths) and In Good Faith (a collection of devotional poems). A memoir of a childhood in the 1930s, For Old Times Sake has been published recently. His academic interests since retirement have been mainly in Anglo-Saxon poetry: a book of translations and commentaries, A Departed Music, has been published, and a second volume is in preparation. He is 82, badly balanced, and blest with a very patient wife. Yeshayahu Shen is Associate Professor in cognitive studies of language in the Department of Poetics & Comparative Literature at Tel Aviv University. His main areas of research are story grammars, discourse comprehension, figurative language comprehension, cognitive poetics, metaphor and conceptual structure, and literary theory. Some of his recent publications appeared in Journal of Memory and Language, Discourse Processes, Cognitive Linguistics, Poetics, TEXT, Journal of Pragmatics, Language and Literature, Journal of Literary Semantics, Empirical Study of the Arts and Poetics Today. He is on the board of Metaphor and Symbol, and is a contributor to the Hebrew Encyclopedia as well as Elsevier’s Encyclopedia of Language and Linguistics.
Index of authors
A Abelson, R.P. 141 Abu-Assad, H. 139, 140 Allan, S. 301 Allen, J. 150, 178 Anderson, B. 77, 114, 318 Andringa, E. 114 Aristotle 338 Astington, J.W. 131 Auracher, J. xii, 268 B Baayen, R.H. 164 Bakhtin, M.M. 23, 24, 28 Balaban, N. 64 Banfield, A. 23, 24, 31 Bard, E. 178 Batelaan, P. 286 Benesh, N. 158, 175 Berlyne, D.E. 230, 234 Biber, D. 162, 163 Black, J. 114, 299, 300, 302, 305, 306 Bogen, J.E. 234 Boonthum, C. 162 Booth, W.C. 14, 114 Borkum, J. 231 Bortolussi, M. xii, 9, 10, 25, 29, 31, 32, 71, 78, 80, 81 Bourdieu, P. 8 Bourg, T. 144 Bower, G.H. 114, 286 Branscombe, N.R. 118 Bray, J. 23, 28, 29, 30, 32 Brewer, M.B. 116 Brock, T.C. 131, 133, 143, 150, 324 Brown, R.J. 116 Bruner, J. 234 Bryant, J. 149 Buckner, R.L. 131 Bunge, M. 3, 10 Burney, F. 29
Burrows, J.F. 211, 212, 213, 214, 215, 217 Busel, V. 287 C Cai, Z. 162, 165 Cantor, J. 150 Carminati, M.N. 83, 84 Carroll, D.C. 73, 131 Cassirer, E. 234 Chafe, W.L. 163, 178 Chambers, E.K. 332 Charness, N. 163 Charniak, E. 165 Chekhov, A. 104, 106, 109, 133, 134, 135, 177, 178, 180, 182, 184, 185 Chesnokova, A. x, xii, 267, 268 Chisango, T. 117 Christianson, S.A. 286 Clark, H.H. 164 Cohn, D. 23 Coleridge, S.T. 58, 100 Collingwood, R.G. 50 Conrad, S. 162 Correll, J. 117 Coulson, S. 59 Crane, R.S. 17 Crossley, S.A. 177, 180, 183 Cupchik, G.C. 77 D Danaher, D. 194 Daniel, F. 162, 165 Darwin, C. 233 Darwin, E. 232 Davis, M.H. 115 Day, L. 117 de Waal, F.B.M. 115 dela Paz, J. 132 Demoulin, S. 117 Dennis, S. 162, 166, 177 Dickinson, E. 40, 186, 188
Dixon, P. 9, 10, 25, 29, 31, 32, 71, 75, 78, 80, 81 Djikic, M. 127, 134 Doan, T. 117 Dolezel, L. 17 Doosje, B. 118 Dovidio, J.F. 117 Downes, W. 194 Du Bois, J.W. 178 Dudusova, M. 268, 309 Dukewich, T.L. 54 Dumais, S.T. 177, 179 Dzyuba, I. 287 E Echols, L.D. 128 Eisenberg, N. 115 Eliot, G. 130, 178 Eliot, T.S. 158, 223, 330 Ellis, J. 37 Ellis, R.D. 52 Emmot, C. 25 Ericsson, K.A. 127, 163 F Fabb, N. 176 Fauconnier, G. 40 Faulstich, W. 114 Fauskanger, H.K. 302, 305 Fellbaum, C. 162, 164, 167 Feltovich, P.J. 163 Fialho, O. 73 Fidler, D.S. 130 Fiedler, K. 117 Fischer, M.H. 83 Flerx, V.C. 130 Fludernik 23, 24 Fortunati, V. 247 Fowler, R. 176, 188 Francis, M.E. 162, 164 Freeman, D.C. 4, 41 45, 194, 201 Freeman, M.H. 40
Index of authors Freud, S. 234, 239 Frith, C.D. 131 Frith, U. 131 Fromm, E. 234, 239 Frost, R. 158 Frye, N. 8 Fülöp, É. 119 G Galinsky, A.D. 117 Garrod, S. 162 Gendlin, E.T. 4, 51, 52, 53, 54, 55, 56, 59 Gerbner, G. 149 Gernsbacher, M.A. 162 Gerrig, R.J. 25, 114, 115, 324 Gibbs, R.W. 82, 194 Giertych, R. 317, 318 Glucksberg, S. 56, 57, 97 Godfrey, J.J. 178 Goh, S.T. 32 Goldman, S. 162 Goldstein, K. 234 Goldstein, T. 151 Goodman, N. 151 Graesser, A.C. 3, 25, 31, 158, 161, 162, 165, 167, 178 Graff, D. 178 Green, M.C. 131, 133, 143, 145, 150, 180, 324 Gregory, W.L. 143 Gross, L. 149 Gross, P.R. 38 H Hakemulder, J. 32, 72, 84, 103, 104, 109, 130, 131, 132, 133, 139, 141, 143, 145, 146, 148, 267 Hall, G. 3, 4, 17, 21, 22, 23, 26, 27, 32 Halliday, M.A.K. 165, 249, 258, 259 Harris, P.L. 131 Harrison, M.R. 128 Hart, M. 186 Hasan, R. 165 Hauptmeier, H. 8 Heal, J. 131 Heeman, P.A. 178 Heidegger, M. 51, 54 Hempfer, K. 17 Herder, J.G. von 318
Hillis-Miller, J. 318 Hirsh, J. 132 Hobbes, T. 234 Hoffman, R.R. 163 Hollander, J. 332 Holliman, E. 178 Hoover, D.L. 158 Hoppe, K. 239 Huizinga, J. 141 Hunt, R.A. 28, 31, 77 I Indurkhya, B. 57 Isen, A.M. 286 Iser, W. 4, 17 Ivanishin, V. 287 J Jakobson, R. 245, 248, 251, 258, 259 James, H. 83, 158, 197, 201, 205, 206 Janis, I.L. 142 Jefferson, A. 175 John, O.P. 117, 134, 180 Johnson, M. 196, 318 Jones, E.E. 114 Jung, C.G. 234 Jurafsky, D. 162, 166, 180 K Kant, I. 4 Keen, S. 115, 116, 123, 130 Kellogg, R.T. 163 Kennedy-Moore, E. 50 Kermode, F. 332 Kernan, A.B. 319 Kim, A.S.N. 132, 309 Kimmel, M. 158, 193, 194, 201 King, B.T. 188 Kintsch, W. 162, 164, 166, 167, 171, 177 Kist, O. 298 Klare, G.R. 164 Klinger, E. 234 Kohlberg, L. 141 Kotsina, I. 286 Köváriné Somogyvári, I. 116 Kozak, M.N. 117 Kreutz, R. 25, 31 Kris, E. 234 Krishnamurthy, R. 245, 261
Kuiken, D. 4, 71, 150 Kuno, S. 114 L Lakoff, G. 194, 196, 206 Landauer, T.K. 162, 166, 177, 179 Lappin, S. 166 Larsen, S.F. 114 László, J. 72, 114, 116, 119, 123 Lawrence, D.H. 23, 26, 27, 28, 31, 330 Leass, H.J. 166 Lee, K.M. 158 Leech, G. 23, 236 Leippe, M. 143 Lenat, D.B. 162 Lentricchia, F. 39 Leung, C. 150 Levinstein, I. 162 Levkovska, N. 285 Lévy–Bruhl, L. 234 Lewis, G. 180, 181, 183, 189 Leyens, J.Ph. 116, 117, 123 Liu, J.H. 116 Lodge, D. 26 Louw, W.E. 159, 176 Louwerse, M.M. xi, 10, 158, 162, 165, 175, 177, 178, 180, 181, 183, 189 Lu, S. 162 Luckmann, T. 321 Ludwig, H.W. 114 M Maat, P. 71, 84, 114, 115, 127, 130 Magliano, J.P. 162, 167 Malinowski, B. 247, 248, 252, 253, 258 Mann, K. 120, 181, 184 Manning, C.D. 180 Mansfield, K. 60, 61, 71, 89, 90, 91, 94, 95, 96, 98, 99, 100 Manstead, A.S.R. 118 Mar, R.A. 72, 127, 130, 131, 132, 133 Margolin, U. 3, 4, 7, 12 Martin, J.H. 162, 166, 180 Martindale, A.E. 239 Martindale, C. 158, 229, 231, 232, 233, 234, 235, 236, 237, 238, 239
Index of authors Maryl, M. 268, 317, 320, 321, 324 Maslow, A.H. 234 Maurenen, A. 260 McCarthy, P.M. 162 McHale, B. 22, 29 McKellar, P. 234 McNamara, D.S. 162, 164, 166, 171, 177 Mead, G.H. 115 Melville, H. 158, 205, 206 Mendes, M. 269 Merleau-Ponty, M. 51, 52, 55 Meyer, C. 178 Miall, D.S. 71, 72, 150, 157 Miles, J. 197, 200, 236 Mill, J.S. 77 Millis, K. 162 Mitchell, H.H. 162 Mitchell, H.R. 128 Moore, K. 231 Morgan, B. 158, 161 Morrison, B. 246 Moskowitz, G.B. 117 Muka ovasjý, J. 55, 89, 106, 230, 233, 273 Mukhtar, S. 145 N Neisser, U. 234 Nietzsche, F. 234 Nisbett, R.W. 114 Noordman, L.G.M. 165 Nussbaum, M. 144 O Oatley, K. 72, 77, 127, 129, 130, 132, 134 Olson, D.R. 131 Owens, J. 114 P Paladino, P.M. 117 Palmer, A. 150 Palmer, F. 141, 142 Panksepp, J. 52 Pascal, R. 23 Pennebaker, J.W. 162, 164 Peterson, J.B. 132, 134 Petrov, V.M. 234 Pfister, M. 17 Phillips, L. 63 Piaget, J. 115, 234
Pichert, J.W. 114 Pickering, M.J. 162 Piepenbrock, R. 164 Pina, A. 117 Podobed, K. 286 Popper, K.R. ix, 9 Postman, L. 77 Powell, B. 128, 309, 310 Prasil, J. 268, 309 Preiss, R.W. 150 Preobrazhenskaya, A. 286 Preston, S.D. 115 Prince, G. 108, 113 R Radevych-Vynnizkiy, Y. 287 Radvansky, G.A. 162, 167 Radway, J. 317, 319, 320, 323, 324, 325 Reinhart, T. 103, 105, 108 Reppen, R. 162 Richards, I.A. 49, 62 Riffaterre, M. 252 Ritterfeld, U. 141 Roberts A.M. 83 Roberts, B.W. 134 Robey, D. 175 Robinson, E.A. 158, 219, 220, 221, 222, 223, 225 Rodríguez, P.A. 117 Rodríguez, T.R. 117 Rogers, R.W. 130 Roitblat, H.L. 32 Ron, M. 25 Rorty, R. 142 Ross, A. 37 Ross, C.S. 134 Ross, P.E. 128 Rowe, J.A. 180 Rusch, G. 8, 9 Russel, R. 117 S Sabine, G. 133 Sabine, P. 133 Sandburg, C. 158, 223, 225 Sanders, T.J.M. 165 de Saussure, F. 8, 36 Saxe, R. 131 Schank, R.C. 141 Schmidt, S.J. xi, 3, 7, 8 Schramm, D. 114 Schreier, M. 27
Schutz, A. 321, 322 Schutze, H. 180 Semin, G.R. 117 Semino, E. 27, 32 Senders, V.L. 77 Sergeyeva, M. 268, 283 Seyffert, P. 16 Shabtai, Y. 106 Shaizhin, A. 286 Shakespeare, W. 4, 35, 42, 43, 44, 178, 181, 245, 331, 334, 335 Shen, Y. 71, 72 Shklovsky, V. (or Šklovskij, V.) 230, 272 Shlonsky, A. 109 Short, M.H. 23, 24, 27, 32 Siddiqui, S. 129 Silkin, J. x Simpson, P. 211 Sinclair, J.M. 176 Singer, M. 167 Slezko, V. 286 Soetaert, R. x Sokal, A. 35ff Sommer, R. 309 Sopcák, P. 55, 85 Sotirova, V. 23, 24, 25, 26, 27, 28, 29, 31, 32 Spears, R. 118 Spreng, R.N. 132 Srivastava, S. 134 Stabler, J. 83 Stanovich, K.E. 128, 129 Steedman.M. 178 Steen, G. 3, 194 Stevens, 158, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 225 Stockwell, P. 194 Stolnitz, J. 53 Stone, P.J. 229, 240 Stoudenets’, G.I. 302 Strange, J. 150, 222, 330 Strayer, J. 115 Stubbs, M. 176 T Taine, H. 233 Takahashi, S. 32 Tannen, D. 163 Taylor, L.J. 162 Tellegen, S. 150
Index of authors Thompson, S.A. 103, 105, 178 Titshall, L. 117 Titzmann, M. 12 Tolkien, J.R.R. 298, 299, 300, 302, 303, 304, 307 Tonkonogy, A. 63 Toolan, M. 23 Tourangeau, R. 57 Trabasso, T. 167 Turner, M. 194, 206 Tynjanov, J. 230, 233 U Uspensky, B.A. 114 V Van der Zee, R. 145, 146 Van Peer, W. xi, xii, 3, 10, 17, 21, 23, 24, 28, 32, 55, 71, 83, 84, 85, 89, 92, 99, 103, 104, 109, 114, 115, 123, 127, 130, 135, 142, 150, 151, 158, 161, 162, 163, 168, 169, 170, 171, 175, 176, 177, 183, 184, 187, 190, 211, 229, 267, 268, 272, 335, 336.
Verbitsky, M. 63 Verdonk, P. 29 Viana, V. x, 159, 267, 268 Vico, G. 234 Viechtbauer, W. 134 Viki, T.G. 117 Vincze, O. 116, 119 Vipond, D. 77 Voloshinov, V. 23 Vorderer, P. 77 W Wagner, R.K. 128 Walton, K.E. 134 Watson, G. 272 Watson, J.C. 50 Weber, J.J. 245 Weber, R. 141 Werner, H. 234, 239 Werth, P. 17, 25 West, A. 239 West, R.F. 128, 129 Wexler, A. 131 Wiebe, J. 113 Winchester, L. 117 Winner, E. 151
Wittgenstein, L. 39, 248, 253, 257, 258, 260, 261 Wolf, W. 17 Wolfe, T. 29 Wu, J. 180, 181, 183, 189 Wundt, W. 234 Y Yepishev, V. 268, 297 Z Zajonc, R.B. 231 Zane, J.P. 177, 180, 186 Zhang, B. 158, 175 Zhirmunskii, V. 11 Zholkovsky, A. 60 Zimbardo, P. 143 Zoeterman, S. 134 Zunshine, L. 130, 150 Zwaan, R.A. 26, 31, 77, 162, 167 Zyngier, S. x, xii, 10, 32, 84, 267, 268
Index of keywords
A absorption 99, 139, 150 aesthetic 50ff, 84, 85, 99, 100, 104, 105, 109, 118, 161, 175, 189, 206, 226, 230ff, 272, 319, 335 see also aesthetic feeling (see feeling) analogue (digital vs. analogue) 243ff ATLAS 193ff B bigram analysis 175, 180ff C canon x, 13, 15, 268, 317ff, 335 (non-)canonical 271ff, 320, 327 changes in personality (see personality change) character(s)(literary character) 22ff, 45, 58, 60, 62, 77, 79, 80ff, 90ff, 109, 113ff, 130ff, 139ff, 184, 188, 201, 212, 236, 303, 308, 141, 143, 144, 145, 146, 148, 150, 151, 180, 202, 268, 338 see also characteristic (also personality, traits) x, 50, 76 cognition 8, 93, 100, 162, 177, 194, 197, 234ff see also cognitive 15, 40, 41, 43, 114, 115, 116, 127, 129, 131, 183, 196, 199, 204, 325 cognitive linguistics (see linguistics) cognitive metaphor (see metaphor) social cognition 114, 131, 139ff
coherence 80, 161ff, 193 cohesion 161ff, 193, 194, 203, 204 Coh-Metrix 161ff collocation 243ff, 274 complexity 4, 21, 24, 44, 51, 57, 59, 79ff, 87, 135, 163, 165, 168, 170, 205, 207, 230, 232, 237, 238, 333 computational 223, 226, 246, 249 see also computational linguistics (see linguistics) computational stylistics (see stylistics) conceptual metaphor (see metaphor) context 7ff, 21, 22, 31, 55, 61, 75, 77, 81, 82, 87, 100, 101, 110, 132, 161, 166, 167, 176, 177, 193ff, 213, 239, 245ff, 273, 309, 311, 321, 329, 335 corpus (corpora) 15, 159, 163, 166, 168, 177ff, 194, 211ff, 246ff corpus analysis 15, 161ff, 208, 212 corpus-attested glossary 159, 243, 259 corpus linguistics (see linguistics) corpus stylistics (see style) correlational analysis 180 cultural studies (see studies) D Deconstructivism see also deconstruction 35ff, 175, 189 radically deconstructed 50
de-lexicalisation (see lexical) deviation 103ff, 238, 251, 271, 272 discourse 7ff, 21ff, 35, 46, 86, 93, 94, 104, 107, 110, 162ff, 189, 194, 195, 200, 271, 274, 336 see also discourse analysis 26, 103ff free indirect discourse 90ff E emotion 30, 41, 51ff, 77, 93ff, 113ff, 131, 134, 141, 171, 197, 202, 209, 232, 253, 286, 311, 317ff see also laws of emotion 89ff empathy 23, 24, 28, 30, 96, 109, 113ff, 127ff, 144, 150, 151 empirical ixff, 7ff, 21ff, 35, 37, 57, 64, 83, 86, 90, 116, 127, 130, 132, 159, 161, 162, 163, 189, 226, 236, 243ff, 278 see also empirical study (see study) empirical stylistics (see style) entropy 229, 232, 238 episode 90, 91, 92, 96, 97, 100 see also episodic structure 89, 91 ethnographic 21, 25, 29, 30, 32 see also ethnographic research (see research) experiment xi, 7ff, 21ff, 75ff, 116, 141, 194, 229, 231 see also experimental manipulation (see manipulation)
Index of keywords expertise 37, 39, 127, 133, 163, 164, 170 see also literary reading expertise 27, 76, 77 psychology of expertise (see psychology) extratextual (see manipulation) F fear 52, 97, 98, 100, 101, 197, 202, 222, 309ff feeling 49ff, 89ff, 115, 119, 197, 310, 311, 320 see also aesthetic feeling 54ff foregrounding xi, 44, 49, 71ff, 78, 89ff, 103ff, 127, 175, 176, 189, 271ff, 346, 348 see also ForegroundBackground 103ff, 272, 273 foregrounded 55, 57, 58, 78, 89 G genre 7, 11, 12, 40, 123, 132, 151, 161, 162, 168, 175ff, 194, 204, 205, 231, 246, 254, 273, 334 see also subgenre 83, 84 H habituation 231ff I IGEL xii, 157, 160, 162, 171 identification 29, 49, 57, 60ff, 72, 81, 98, 109, 113ff, 150 image schemata 193ff imagination 50, 55, 139ff, 216, 220, 225 see also imagining scenario 141 inference(s) 76, 83, 85, 86 infrahumanization 113, 116ff interdisciplinary x, xi, 35ff, 36, 38, 39, 40, 161 Iota 211ff
L language 9, 23ff, 36, 40ff, 84, 86, 96, 98, 103, 104, 108, 110, 128, 129, 140, 161ff, 176ff, 200, 205, 213, 216, 244ff, 272ff, 282ff, 297ff, 312, 329, 337, 340 see also language learning 249, 283ff Latent Semantic Analysis (LSA) 166, 170, 175ff laws of emotion (see emotion) lexical 24, 104, 166, 222, 297, 298, 300 see also de- lexicalisation 243ff lexical bundles 271ff lexical density 271, 273 lexical items 180ff lexical variety 268, 271ff lexicography 243, 248 re-lexicalisation 243ff linguistics x, xi, 7, 16, 22ff, 36ff, 103ff, 117, 120ff, 162, 163, 176, 180, 183, 189, 196, 205, 230, 238, 243, 248ff, 271ff, 286, 287, 297, 307 see also cognitive linguistics 193ff computational linguistics xi, 161ff, 175ff corpus linguistics 195, 271ff linguistic inter-group bias 113ff literary ixff, 7ff, 21ff, 35ff, 55, 59, 60, 64, 83, 89ff, 103ff, 114, 115, 130, 133, 134, 135, 143, 147, 150, 151, 161, 162, 171ff, 193ff, 212, 226, 229, 231, 237, 239, 243ff, 271ff, 317ff see also literariness 127, 150, 175, 189, 208, 268, 271ff literary communication 7ff literary narratives (see narrative) literary reading 21ff, 49ff, 151, 326 literary reading expertise (see expertise)
literary response (see response) Literary Response Questionnaire (LRQ) 150 literary study (also: study of literature) ix, x, 8, 35ff, 146, 175, 183, 271, 272, 273 literary theory (see theory) M manipulation 8, 15, 75ff, 197, 200, 272, 273 see also experimental manipulation 75, 78, 80, 81, 83ff extratextual manipulation 86 mental models 25, 161ff metaphor 11, 24, 35ff, 51ff, 78, 96, 110, 158, 159, 193ff, 235, 247, 255, 261, 281 see also cognitive metaphor 35ff conceptual metaphor 193ff methodology ix, 7, 10, 11, 18, 22, 27, 28, 243, 247ff motivation 81, 93, 283ff, 323, 325 N narrative x, 10, 11, 16, 17, 23, 24, 37, 39, 43, 58, 82, 89, 90, 92, 103, 105, 113ff, 130ff, 139ff, 168, 187, 189, 194, 197, 201, 202, 209, 222, 234, 239, 272, 275 literary narratives 103ff, 114, 115, 151, 194, 239 narrative perspective (see perspective) non-literary 15, 158, 161, 162, 175ff, 193, 325 novelty 229ff, 272, 334 P participatory response (see response) perception 32, 51, 52, 79, 92, 93, 100, 116, 141ff, 162, 211, 230, 272, 308, 309, 310, 311, 314
Index of keywords personality 80, 134, 135, 347 see also personality change (also: changes in personality) 127ff personality measures 76, 134 personality traits 79, 132, 134 transformation of personality 347 perspective (reader’s) 25, 27, 30, 49, 92ff, 114ff, 165, 197, 207, 208, 320 see also narrative perspective 72, 89, 113ff, 209 phenomenology 49ff phonetic 89, 94, 95, 96, 248, 297ff poetry 17, 45, 62, 83, 84, 211ff, 229ff, 246, 273, 329, 330 see also British poetry 229ff history of poetry 229ff popular fiction 150, 317ff post-processing 7, 11, 14, 30 psychology v, ix, x, xi, xii, 3, 8, 15, 17, 33, 66, 68, 71, 87, 116, 124, 127, 129, 130, 186 see also cognitive psychology 127, 347 clinical psychology 345 neuropsychology 151 psychology of expertise 129, 130 social psychology 116 Q qualitative 21, 30, 60, 193, 197, 202, 207, 208, 225, 238, 277, 322 qualitative analysis (also research) 158, 193, 195ff qualitative research 29 quantitative x, 24, 27, 132, 168, 196, 206, 207, 209, 225, 236, 237, 289, 322 R reception 7, 10, 11, 12, 14, 31, 33, 193, 323, 348 see also reception of film, 346
REDES x, xii, 211, 265, 268, 269 relation (causal) 76ff see also semantic relation 176ff interrelation 8 re-lexicalisation (see lexical) research ethnographic research 25 research paradigms 21 qualitative research (see qualitative analysis) response ix, 23ff, 51, 56ff, 75ff, 89ff, 114, 115, 130, 140, 141, 143, 150, 209, 229, 289, 321, 323, 325 literary response 24, 75ff, 89, 325, 347 participatory response 72, 114 S science studies (see study) semantic semantic prosody 243ff, 274 semantic relation (see relation) simulations 15, 98, 127ff, 237 social see also social cognition (see cognition) social interaction 127, 132 social socio-cultural 8 sociological xi, 322, 326 sociology ix, 16, 39, 317, 319, 321, 322 study(ies) see also cultural study(ies) 35ff empirical study(ies) ixff, 7ff, 32, 60, 75, 83, 89, 127, 130, 135, 151, 161, 171, 175, 176, 209, 229, 251, 271, 283 science study(ies) 35ff Society for the Empirical Study of Literature (IGEL) xi, 161 study(ies) of literature ix, xii, 7ff, 35, 46, 84, 127, 151, 162, 175, 229 style 7, 11, 25, 27, 82ff, 107, 117, 135, 161, 164, 180, 206, 209,
211ff, 229, 235, 236, 239, 240, 331, 334 see also empirical stylistics 23 computational stylistics (also data-assisted) 211ff, 243ff corpus stylistics 243ff stylistics xi, 23, 29, 35, 78, 89, 92, 162, 163, 175, 176, 194, 204, 211, 213, 223, 226, 234, 236ff, 243, 247ff, 272, 273 sympathy 29, 77, 98, 114, 115, 127, 130 syntax 46, 109, 161ff, 243ff T tagging 193, 223 theory see also literary theory 7ff, 35ff, 103, 108, 209 theory of mind 127, 131, 139ff traits (also characteristic, personality) 79, 132, 134, 162, 230, 244 translation 12, 180, 248, 297ff transliteration 297ff transportation 55, 139ff, 324, 325 trope 64, 193, 196, 201 type/token 8, 167, 170, 218, 271ff V Variables see also contextual variables 76, 99 dependent variable 95 independent variable 75, 76, 78, 84, 86, 121 reader variables 76, 82, 87 visualization 211 W WordNet 167 Z Zeta 158, 211, 213ff
In the series Linguistic Approaches to Literature the following titles have been published thus far or are scheduled for publication: 5 4 3 2 1
Zyngier, Sonia, Marisa Bortolussi, Anna Chesnokova and Jan Auracher (eds.): Directions in Empirical Literary Studies. In honor of Willie van Peer. 2008. xii, 357 pp. Peer, Willie van (ed.): The Quality of Literature. Linguistic studies in literary evaluation. 2008. ix, 243 pp. McIntyre, Dan: Point of View in Plays. A cognitive stylistic approach to viewpoint in drama and other text-types. 2006. xii, 203 pp. Simpson, Paul: On the Discourse of Satire. Towards a stylistic model of satirical humour. 2003. xiv, 242 pp. Semino, Elena and Jonathan Culpeper (eds.): Cognitive Stylistics. Language and cognition in text analysis. 2002. xvi, 333 pp.