ABSTRACT Title of Document: LITERARY JOINT ATENTION: SOCIAL COGNITION AND THE PUZLES OF MODERNISM Vera L. Tobin, Ph.D., 2008 Directed By: Profesor Mark Turner, Department of English The fundamental claim of this project is that the mechanics of social cognition?how we think intersubjectively and proces social information?are highly relevant to the study of literature. Specificaly, it presents a theory of literary discourse as the emergent product of a network of joint activities and joint atention. Research on joint atention frequently focuses on contexts in which this aspect of social cognition is not fully developed, as in autism and early childhood. The study of literature, on the other hand, is continualy engaged with circumstances in which joint atention is relevant, highly developed, and complex. Here, linguistics and cognitive science provide the basis for specific and particularizing claims about literature, while literary texts are used to support broader theoretical work about language and the mind. The focus is on modern literature in English and its reception. Many of these texts exploit systematic egocentric biases in social cognition and communication to produce efective ironies and narative surprises. Further, both detective fiction and experimental Modernist fiction frequently dramatize problems of joint atention that can be traced to the ultimate relation betwen author, reader, and text. Extended analysis, with special atention to Edgar Alan Poe?s ?Murders in the Rue Morgue? and Virginia Woolf?s To the Lighthouse, demonstrates the importance of this joint atentional trope. In these texts, the external and perceptible serve not only as triggers for the events of a single consciousnes, but as a locus for the potential for intersubjective experience, both inside and outside the text. A case study of the publication and reception history of Marianne Moore?s ?Poetry,? finaly, demonstrates the utility of a cognitively realistic approach to textual criticism. These literary activities also serve as an important proving ground for the claims of cognitive science, demonstrating complexities of and constraints on shared viewpoint phenomena. . LITERARY JOINT ATENTION: SOCIAL COGNITION AND THE PUZLES OF MODERNISM By Vera L. Tobin Disertation submited to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfilment of the requirements for the degre of Doctor of Philosophy 2008 Advisory Commite: Profesor Mark Turner, Chair Profesor Eve Swetser Asociate Profesor Michael Israel Asociate Profesor Peter Malios Asociate Profesor Jefrey Lidz ? Copyright by Vera L. Tobin 2008 i Dedication For Steve. ii Table of Contents Dedication.........................................................i Table of Contents...................................................ii List of Tables......................................................v List of Figures.....................................................vi Chapter 1: Introduction...............................................1 Literary Joint Atention.............................................1 Overview of Chapters..............................................3 Chapter 2: Preliminaries..............................................8 Introduction......................................................8 Reading Minds...................................................8 Common Ground.................................................13 Shared Basis....................................................19 Atention.......................................................21 Joint Atention...................................................24 What It Isn?t..................................................24 hat It Is....................................................25 Intentions......................................................28 Developmental Timeline...........................................33 Chapter 3: Participants in the Literary Event..............................43 From Conversation to Literature.....................................43 Readers, Authors, and ?The Intentional Falacy?.........................44 Intention and Meaning.............................................51 Theoretical Commitments: Language Use is a Joint Activity................53 The General Idea...............................................53 Application to Textual Discourse...................................59 Theoretical Commitments I: On the Nature of ?Texts? and ?Discourses?......61 Text.........................................................61 Discourse....................................................63 Common Objections to the Conversation Model for Texts..................66 Authors and Readers are not Physicaly Co-Present.....................66 Who Is Speaking?..............................................67 Proposal: The Literary Event Consists of Multiple Joint Activities...........70 Inter-Reader Interactions...........................................71 Collaboration in Conversation.....................................71 Collaboration in Text Interpretation.................................73 Collaboration and Change Within Reading Communities................76 Back to Readers and Authors........................................81 Summary.......................................................84 Chapter 4: Efects of Egocentrism......................................86 Social Cognition and Literary Criticism................................86 The Curse of Knowledge...........................................92 Shifting Viewpoints...............................................97 iv Zooming In: Imersive Fiction and the Narative Rug-Pull................105 The Twist Ending.............................................105 The Murder of Roger Ackroyd...................................109 Zooming Out: Irony and the Curse of Knowledge.......................113 Summary......................................................124 Chapter 5: Puzzles of Joint Atention in Poe and Woolf....................127 New Readings..................................................127 The Modernists and the Detectives..................................129 Reading Minds in the ?Rue Morgue?.................................134 In To the Lighthouse.............................................144 Shared Consciousnes and Joint Atention...........................144 The Reader as Dupin...........................................148 A Kitchen Table When Only You Are There.........................154 Summary......................................................159 Chapter 6: The Editorial Conversation.................................162 Criticism in Practice.............................................162 Editing and Conversation.........................................165 Revision and ?Poetry?............................................166 Omisions Are Not Acidents (But They Aren?t the Whole Story, Either).....170 Joint Construal and Revision in Conversation..........................180 Eclectic Texts and Social Texts.....................................185 Thematizing Trouble, Thematizing Repair.............................188 Summary......................................................192 Chapter 7: Conclusion..............................................194 Bibliography.....................................................196 v List of Tables Table 5-1: A summarized edition history of ?Poetry????????????.. 172 vi List of Figures Figure 1-1: Joint atention and symbolic communication??????????.. 26 Figure 3-1: Scope of awarenes and common ground in Mark and June narative.. 99 Figure 3-2: A narative configuration of mental spaces??????????? 102 Figure 3-3: Zooming out???????????????????..??? 116 Figure 3-4: ?Lovely weather?????????????????????? 117 Figure 3-5: Clashing models in Treasure Seekers?????????????. 121 Figure 3-6: ?Thy ready wit????????????????????.??.. 123 1 Chapter 1: Introduction Literary Joint Atention One of the goals of cognitive linguistics, broadly speaking, is to investigate what language reveals about the mind: not what one person?s language reveals about her own mind, but what paterns and universals of linguistic phenomena reveal about the workings of human minds in general. Scholars of literature tend to have very diferent goals, investigating historical, political, and symbolic paterns in order to cast light on particular works of literature, authors, audiences, and genres. But any complete theory of language and the mind should be able to acount for what is going on when a person reads a novel. This is a daunting prospect, to say the least: the number of things that are ?going on? is imense and litle of it is easily acesible through direct observation. This project is an atempt to present an acount of literary communication that fits in with existing usage-based acounts of face-to-face discourse, and furthermore to demonstrate that such an acount can yield rich new insights into individual texts. Holding a conversation would sem as if it ought to demand a great deal more work than producing or understanding a monologue: Conversation involves coping with unpredictable twists introduced by your conversational partner, making sure that what you say is presented in a way that takes your partner into acount, and dealing with fragmentary and eliptical utterances, al very rapidly. Yet most people find it 2 much more dificult to deliver or even listen to a lecture?let alone to read one?than to cary on a conversation. This observation, among others, suggests that that humans or language, or both, are optimized for thinking in interaction with each other. I take the position that conversation is a joint activity (Clark 1996) that capitalizes on our special abilities of social cognition. A fundamental underpinning of this kind of communicative coordination is the ability to engage in sustained scenes of joint atention. Some time around their ninth to twelfth months of life, children begin to engage in a number of new activities that involve not just themselves and an object nor just themselves and another person, but themselves, an adult, and objects in their environment toward which both infant and adult direct their atention. Before this age, infants engage in dyadic behaviors? interactions betwen themselves and an adult, or themselves and an object?but around their first birthday, they begin to flexibly and reliably coordinate their activities within a referential triangle of themselves, an adult, and an object or event towards which they're sharing atention. This ability sems to be a crucial ingredient in language acquisition. Early signs of linguistic communication co-occur with the emergence of these joint atentional activities; some findings indicate that the amount of time spent in joint atention activities in the first year is predictive of vocabulary at 18 months. 1 Researchers who study joint atention frequently apply the framework to cases such as autism and early childhood, in which people?s ability to understand 1 For reviews of the many studies that support these claims, se Carpenter, Nagel, and Tomaselo 1998 and Tomaselo 2000. The specific correlation betwen frequency and duration of joint atentional behaviors and childhood vocabulary is discussed in detail in Tomaselo and Todd 1983. 3 themselves as part of these scenes is not fully developed. The study of literature, on the other hand, is continualy engaged with circumstances where joint atention is relevant, highly developed, and complex. In this study, linguistics and cognitive science provide the basis for specific and particularizing claims about literature, while literary texts are used to support broader theoretical work about language and the mind. The fundamental claim of this project is that social cognition?how we think intersubjectively and proces social information?is highly relevant to the study of literature. Theoretical analyses of discourse procesing often asume that texts in general and published fictional naratives in particular should be thought of as a form of interaction betwen the author and the reader, in which writers and readers occupy the positions of speaker and addrese, while many theorists engaged in the humanistic study of texts object to treating them as examples of communication at al. This study presents a theory of literary discourse as the emergent product of a large and complicated network of joint activities and scenes of joint atention. This theory then generates concrete and specific insights about textuality, the nature of participation in literary discourse, and the interpretation of individual texts. Overview of Chapters In its broad structure, the disertation is organized as follows. In the folowing chapter, ?Preliminaries,? I introduce and define several crucial technical concepts, particularly atention, joint atention, communicative intention, and joint intentions. Next, I consider the question of who, if anyone, in the literary situation can properly 4 be understood as involved in any kind of ?joint? engagement with one another. This discussion addreses a number of clasical theoretical objections to treating literary texts as examples of communication or speculating about the communicative intentions of authors and readers. I argue that these models of language use, which separate intention and meaning, are not satisfactory, and instead endorse a model in which language use is fundamentaly an activity performed jointly by participants in a discourse. This theory, based primarily on the work of Clark (1996) and Lewis (1969), is here explained in detail. The remainder of chapter thre is an argument for treating literature as a complex of multiple joint activities, some fictive and some not. I argue that there are at least four kinds of intersecting joint activities involved in the discourse situation of a published narative text, each of which is discussed in turn: (1) represented scenes of joint atention that take place betwen characters; (2) interactions among participants in the development and production of the published text: the credited author, editors, publishers, and others; (3) the relationship betwen historical authors and individual readers, sometimes mediated by their fictive counterparts; and (4) interactions in which texts themselves serve as objects of joint atention within reading communities. Chapter four takes a closer look at the details of social cognition, including how, when, and how acurately we conceptualize the minds of others, and argues that these mechanics are of direct interest to literary studies. I claim that literature is in an important sense made out of scenes of joint atention, both in practice (the activities of the many real people involved in producing, reading, distributing, and responding 5 to texts) and in representation (the activities of literary characters and the encoding and manipulation of viewpoint in naration). To the extent that this is so, literature is bound by the rules and mechanisms of joint atention and, by extension, of social cognition?including biases in the system. Here, I concentrate on just one such bias, the ?curse of knowledge,? and show how it can give rise to two kinds of literary efects: On the one hand, texts can take advantage of readers? tendency to align their own viewpoints with a represented embedded viewpoint, and then to fal prey to the curse of knowledge, failing to discount this additional information when imagining what others think. Texts that exploit this tendency can use it to engineer ?rug-pul? endings, surprising readers with information that contradicts the over-generalized propositions. On the other hand, some texts prompt readers to recognize characters? susceptibility to the curse of knowledge, showing them to be poor judges of the transparency of their own communicative intentions. The discrepancy betwen the way the characters understand their own situation and what the reader is given to understand the true situation to be is a species of dramatic irony. In recognizing the source of this discrepancy, readers ?zoom out? to a more distant perspective on the characters? perspectives, creating a sense of ironic distance and complicity with an implied author. Examples discussed in this chapter include Jane Austen?s Ema, E. Nesbit?s children?s novel The Story of the Treasure Seekers, and Agatha Christie?s The Murder of Roger Ackroyd. The second part of this study presents case studies of how work on joint atention and social cognition can open the door to new ays of reading individual 6 texts. These readings demonstrate, I hope, that this approach generates specific insights of interest to traditional literary criticism. Chapter five discusses the thematization of and structural reliance on scenes of joint atention to be found in both clasic detective fiction, especialy Edgar Alan Poe?s ?Murders in the Rue Morgue,? and British modernism, particularly Virginia Woolf?s To the Lighthouse. This analysis highlights the ways in which naratives of detection can in an important sense be ?stream of consciousnes? naratives themselves. It also reveals a number of peculiar isomorphisms betwen these genres? approaches to the relationship betwen intersubjectivity, meaning, and the perceptible physical world. In both cases it turns out that perceptibility itself is les important than shared perception, and that in both it is the quality of shared consciousnes that gives our consciousnes our shape. In my final chapter, I provide a social-cognitive analysis of the stylistics and textual history of one of the most famously revised poems in modern English, Marianne Moore?s ?Poetry.? I argue that the thematization of omision in ?Poetry? is best analyzed as a joint construal both of the poem and of its history, built up together by author, editors, publishers, critics, and other readers. This framework helps to reconcile some problems that plague both the ?final authorial intention? approach to textual criticism, asociated with W. . Greg, Fredson Bowers, and G. Thomas Tansele, and the ?social text? approach, most famously asociated with Jerome McGann. The social text approach is philosophicaly appealing, in that it captures the complicated nature of authority and textual construction. It suffers, however, from its atempt to legislate a rationalist, anti-intentionalist approach to editing and reading. 7 Many of the people involved in the socialy situated, socialy constructed production and contextualization of a text wil themselves be relying on intentionalist strategies of interpretation. A cognitively realistic approach that acounts for the real-world behavior and beliefs of readers, editors, and archivists serves to bridge the two approaches, and the mechanics of joint atention provide a productive framework for talking about these proceses without obscuring either aspect of the facts on the ground. It also provides an opportunity to look at the longitudinal stylistics of a literary work as a discourse unfolding over time, and iluminates the connections betwen these literary acts and their counterparts in ordinary conversation. 8 Chapter 2: Preliminaries Introduction This chapter presents an introductory overview of concepts and phenomena that wil be central to the analyses to follow. This project is an argument for the importance of joint atention and joint activities to the study of literary discourse; before I begin, then, I should at the very least provide the reader with an explanation of what these phenomena are, a sense of why they are important for things other than literary discourse, and what they have to do with language. Because these questions center on formulating shared communicative intentions and monitoring what we know and intend in common with other people, an introduction to joint atention must begin with an introduction to the kinds of problems that joint atention sems to solve. Reading Minds Humans are experts at understanding the behavior of themselves and others in terms of beliefs, desires, and intentions. We can ascribe mental states (which we recognize may difer from our own) to other people; we can predict how these mental states might change in response to various things; and we can use these ascribed mental states to predict or make sense of behavior. Al of this comprises our everyday ?mind reading? (Baron-Cohen 1995), sometimes caled ?theory of mind? (Premack 9 and Woodruff 1978) or ?concept of mind?. Because we don?t have the advantage of direct telepathic aces to one another?s thoughts, we have to make do with inferences based on a combination of our background knowledge and our observations of what people say and do. Autistic people have a lot of trouble with this; indeed, many clinicians treat impairment of this ability as a central characteristic of autistic spectrum disorders. If you?re a neurotypical adult, however, you do this kind of thinking automaticaly, al the time. It sems that most of us can?t help but think of other people in this way, any more than we can help recognizing a word in a language we know (cf. Stroop 1935). The mechanism of this competence, however, is a somewhat contentious subject. First, there is significant disagrement over the question of when and how infants or children develop a ?real? understanding of other people as psychological beings with mental states that may difer from their own. Meltzoff (1995) proposes that research on social-cognitive development should be understood as testing two diferent questions. On the one hand, we can investigate the development of mentalism: how and when children first begin to construe others as having psychological states that underlie their behavior. On the other, we can investigate the development of a representational model of other minds, that is, when children come to be able to formulate not just implicit or operational, but explicit representations of other people?s atitudes and beliefs, and when they understand mental states as active interpretations of the world. The most stringent tests for social understanding concentrate on the later of these two abilities, and most suggest that children acquire a representational theory of 10 mind sometime around their fourth or fifth year. The most famous of these tests is the false belief task (Wimer and Perner 1983; Leslie and Frith 1988), which tests whether or not children wil atribute false beliefs to others. In this task, the child is presented with a scene in which another person (let?s cal her Ana) ses a set of objects in a particular configuration and then leaves the room. While she?s out of the room, the situation changes; for example, a marble may be moved from a closed basket to a closed box. The child is asked what wil happen when Ana returns to the room: where wil she look for the marble? To pas the test, the child must understand that Ana does not know what the child knows. Typicaly developing children under the age of four and most autistic children fail this test, wrongly claiming that the other person wil look for the object in the second location, where the child knows it realy is. Similar age efects are observed in tasks testing related skils, such as appearance- reality (Flavel, Gren, and Flavel 1986), in which children are shown something like a familiar candy box that is revealed to be filed with something other than candy, and asked both ?What does it look like is in the box?? and ?What?s realy in it?? Again, thre year olds consistently fail this kind of task, while non-autistic five year olds consistently succed. There is, however, some evidence that sems to complicate this developmental timeline. For example, Onishi and Bailargeon (2005) developed a nonverbal false belief task which children as young as fiften months old sem to pas. In this test, infants were made familiar with a scene in which an adult actor hides and then retrieves a toy in one of two boxes. In the trials, then, the actor first hides the toy in one of the boxes. In the false belief condition, the actor then leaves 11 the scene. In the true-belief condition, the actor stays in the room, watching, as the object moves by itself from the original box to the other. Infants? apparent expectations about where the actor would search for the toy were tested by measuring their looking time under a variety of conditions, and the infants did indeed appear to expect the actor to look for the toy based on the actor?s belief rather than the toy?s actual location. Indeed, in general, looking-time studies with infants and other tests of manifest rather than explicit social-cognitive skils suggest that at least some kind of implicit mentalist thinking arises a fair time before children can succed at traditional false-belief tasks that require them to respond to explicit questions about the beliefs of others. Woodward (1998), for example, uses the looking-time paradigm to show that before even their first birthday, infants focus selectively on aspects of behavior that are relevant to the actor?s underlying intentions, examining grasping and non- grasping reaching events. Meltzoff (1995) showed 18-month-old children conditions in which an adult either achieved some result on an object or tried but failed to achieve the same result and tested whether they would re-enact what the adult did or the adult?s intended action. The infants produced the completed result as often in the try condition as in the succes condition, suggesting that they saw the adult?s behavior as goal-oriented in both cases. (Notably, they did not produce the target acts for a ?try? condition when the actor was replaced by a mechanical device.) Thus, even infants sem to understand, at least implicitly, quite a lot about behavior, and can make use of their observations of intentional behavior in many subtle and sophisticated ways. Whether this understanding includes any kind of 12 model of beliefs is les clear. It appears that understanding of intentions, especialy implicit understanding of intention, arises much earlier than explicit reasoning about others? beliefs. This is also the level of social cognition that appears to be necesary for early language acquisition (se, for example, Tomaselo et al 2004). Another major line of disagrement (and there are many diferences of opinion within these broad categories as wel) in this area is betwen those who think that these inferences are made by reference to a real theory of other minds and those who think it relies primarily on running simulations of the mental proceses we ascribe. The first group of researchers argue for the so-caled ?theory theory? (e.g. Morton 1980; Gopnik and Meltzoff 1997; Gopnik and Welman 1994). This view holds that social understanding relies on reference to an internaly held theory of what other minds consist of and what they can do. A prevailing version of this acount argues that children develop their understanding of other minds much in the same way that scientists develop theories about other phenomena. That is, they generate abstract, coherent hypotheses and make predictions based on them. Then they test those predictions, gathering evidence through experimental explorations of the world. Sometimes this evidence confirms their theory; sometimes it is at odds with the theory, which can lead the child either to reinterpret the results in terms of their working theory or to sek a new theory that beter fits the evidence. In the theory theory, then, understanding the thoughts and intentions of others involves an explicit representation of other people?s mental states, based on an actual theory of how these other minds operate. This theory begins with a simple model and grows more complex and refined over time. 13 The primary alternative to this acount is the simulation theory (e.g. Goldman 1992; Haris 1992). This view holds that when we interpret, explain, or make predictions about other people and their mental states, we do it not by reference to an internaly represented theory but by simulating similar proceses in ourselves. This acount, some argue (e.g. Galese 2007), acords with the large body of research on the role of simulation in people?s understanding of other people?s afective experience?that is, what happens when we watch, for example, someone else experiencing pain (Jackson et al 2005) or disgust (Wicker et al 2003). Perceiving and asesing these felings in others is asociated with activation of some of the same neural structures asociated with the direct experience of that emotion. The claim in the simulation acount of social cognition is that something very similar, not any reference to a general ?theory? of mind, is what lies behind thinking about others? beliefs, goals, and other intentional states. Stil, most researchers can at least agre that people realy do form inferences about other people?s mental states. One of the things that we can do with this inferencing is to formulate an idea?implicit or explicit?of what is common ground betwen ourselves and the people with whom we interact. Common Ground Common ground refers to the knowledge, beliefs, and atitudes that interlocutors not only share but also recognize that they share. The discussion of common ground to follow is largely derived from the work of Clark (1996). This technical notion of common ground was first proposed by the philosopher Robert 14 Stalnaker as part of his discussion of the linguistic phenomenon of presupposition (1978). Stalnaker, arguing against semantic theories of presupposition in sentences like ?My mother is tal? (which presupposes that the speaker has a mother), proposed a definition of presupposition that centers on speakers? asesment of common ground. ?To presuppose something is to take it for granted, or at least to act as if one takes it for granted, as background information?as common ground among the participants in the conversation? (Stalnaker 2002:701). Similarly, if the presupposed content was not in fact already part of the common ground, a presupposition trigger wil (in most cases) prompt hearers to accommodate the presupposition by adding it to the common ground. This notion of common ground has its roots in older notions such as common knowledge (Lewis 1969) and undefined uses of the term in Grice (1967). The position that I adopt in this study is that our asesment of what is or is not part of our common ground determines a great deal of what we choose to expres and how we choose to expres it. 2 The simplest example of how our asesment of common ground afects our linguistic choices is in the use of definite and indefinite articles (Clark and Marshal 1981). 2 This claim, especialy in the very volitional form I have stated it here, has its detractors. For example, there is plenty of room for debate about the amount and type of recipient design that is actualy performed on-line in the quick, demanding circumstances of face-to-face conversation (c.f. Keysar et al. 1998). Thoughts difer as wel on the question of how much of language production is mechanistic and pre- fabricated (c.f. Pickering and Garod 2004). It is possible that many of these adjustments are so highly automated that a detailed mental model of common ground plays litle part in their construction. Nonetheles, people?s behavior correlates to a substantial degre with expected acommodation of what they have reason to believe is and is not common ground. I wil discuss some important divergences from this patern in chapter thre. 15 (1) a. Have you read the book about squirels? b. Have you sen a book about squirels? (1a) construes the book in question as part of the interlocutors? common ground, while (1b) does not. But concerns of common ground shape our linguistic choices wel beyond the choice betwen definite and indefinite reference. Clausal structure in general reflects a slew of choices of this kind that are often described under the umbrela of information structure (Chafe 1976, Lambrecht 1994). Gramatical structures can reflect common-ground considerations such as whether an argument is discourse-old (given) or discourse-new (Prince 1981) or the need to establish focus on a particular entity or event (Lambrecht 1994). For example, cleft constructions, left-dislocation constructions, and topicalization constructions are al asociated with diferent information-structural properties. Further, considerations of common ground are reflected in language-level conversational choices such as what language to speak in, which register to adopt, and other such considerations. If I make an off-color joke during a conversation, that choice wil hinge in part on my asesment of what my interlocutor and I mutualy understand to be the nature of our relationship and the seting and type of interaction we mutualy understand ourselves to be engaging in; it may also serve in part as an atempt to shape that mutual understanding. It may also, of course, misfire badly. When I am in Italy, it usualy is a safe gues that a stranger on the stret speaks Italian, but as a foreigner I wil keep my eye out for signs one way or the other. If I want to speak in English, I wil try to determine whether my interlocutor, like myself, is a member of the community of English speakers. (This may, of course, turn out to 16 be a mater of degre.) There are some conventional features of English that are common ground to al speakers of English, of any variety (se Lewis 1969 for a definition of linguistic conventions as common knowledge within a community of speakers; I discuss this in more detail in chapter 3). But there are also many lexical items whose use and meaning are conventional only within much smaler communities. Wil it be appropriate and communicative for me to use the word caesura in conversation with Philip? Only if he is an enthusiast of the study of technical properties of scansion and prosody. Clark (1996) identifies two kinds of common ground: communal and personal. Communal common ground derives from inferences based on evidence about the cultural communities, large or smal, that people belong to. It includes beliefs and asumptions about norms of behavior, background knowledge, cultural frames, linguistic conventions and conventions for al kinds of other behaviors and situations, skil competencies, and more. If I find out that my interlocutor Beth is an English-speaking profesor of eightenth-century Portuguese literature living in Washington DC, I have identified her as a member of at least thre diferent communities: the community of speakers of English, the community of profesors of eightenth-century Portuguese literature, the community of residents of Washington, DC. I can now make certain asumptions about what kinds of knowledge and expertise that I can expect her to have, based on her membership in these communities. I wil expect that she knows basic features of the geography of Washington, for example, and that she has tacit knowledge of basic English vocabulary, phonology, gramar, and usage. Since I too am a member of these 17 communities, I can frely atribute the asociated encyclopedias of information to our common ground and proced acordingly, addresing her in English, for example, and giving directions that start at the Dupont Circle Metro station. By contrast, I am not a profesor of, expert in, or enthusiast of eightenth- century Portuguese literature. If Beth and I establish the mutual belief that she is a member of the community of experts in this field and I am not, we are licensed to make a diferent set of asumptions about what is common ground betwen us. Where in the previous cases Beth and I had reason to take as common ground al ?inside? (Clark 1996:101) information of the communities in question?al the information that members of a community asume to be possesed by members of that community?here we can only take ?outside? information to be part of our common ground. Outside information is information about the types of information that people outside a community asume is inside information for members of that community; inside information is not just a mater of types but particulars. If I?m an outsider, Beth and I can mutually asume only that Beth knows the basic landmarks of the city she lives in. If I?m an insider, we can mutualy asume that al manner of specific facts about those specific landmarks are part of our common ground. We also make a number of asumptions about what is common ground betwen ourselves and other people based on the very broadest of community memberships, our membership in the human race and human society. I draw on my folk psychology or theory of mind for people in general to asume, absent evidence to the contrary, that other people have roughly the same kinds of sense experiences and mental states that I do. If, for example, something is visible to me, I expect that it wil 18 be visible to other people in the same circumstances. I asume too (though I may be persuaded that this asumption was incorrect) that what is perceptualy salient to me wil similarly grab the atention of another person in the same circumstance: I wil be surprised if you fail to notice the sound of a loud explosion. I asume that you share my familiarity with the basic laws of nature and biology?that we are mutualy aware of gravity, of the need to eat and drink to sustain life, and mutualy familiar with the experience of these necesities. I asume, too, that certain social universals are common ground betwen myself and wel-nigh any person I met: that we al take as given that people use language, have names and kinship relations, and so on. Personal common ground is the common ground that arises from the personal experiences we share with another person or people. When my friend and I go to the museum and look at a painting together, or have a conversation about the time her mother learned to juggle, we are justified in infering that what we saw or talked about is now part of our common ground. So too is the knowledge of our shared activity. Over the course of a conversation, the things that we say enter our common ground. Things we have said to one another, done together, and experienced together in the past are also part of our personal common ground. Our mutualy perceptible physical surroundings, too, are an important part of our common ground. Making use of these experiences as the basis for our asesments of common ground relies on communal common ground, and vice versa. I have to make asumptions about our common sensory capacities and atentional tendencies if I am to believe that I have any evidence about what we have sen or heard in common. I have to take a slew of 19 linguistic conventions and background information as common ground if I am to believe that you and I mutualy understand your having said ?My sister went outside? to mean that you have aserted that your sister Mary recently walked out of your house. This brings us to the question of how, indeed, people do determine that a piece of information is in the common ground. This wil turn out to hinge importantly on the mechanics of joint intention. Shared Basis A common objection (e.g. Sperber and Wilson 1986) to the principle of common ground or mutual knowledge is the problem of infinite regres, the so-caled ?Mutual Knowledge Paradox?. 3 One way of defining common ground is as a series of iterated propositions: p is common ground for members of C if and only if: 1. members of C have information that p, 2. members of C have information that members of C have information that p, 3. members of C have information that members of C have information that members of C have information that p, and so on, ad infinitum. (Clark 1994:95) The trouble is that this model clearly cannot acurately represent anyone?s actual mental state, because none of us have the infinite procesing or storage capacity to run through the infinite regres of iterations. 3 For a fuller history of this debate, se Le 2001. 20 One possible way of dealing with this problem is to propose that the above acurately describes the logical definition of common ground, while people?s real psychological representation only runs through a few such iterations and declares itself satisfied?rather than actualy running on ad infinitum, we simply form a representation of something like ?and so on, ad infinitum,? and stop there. The main argument against these limited regresion proposals, as sen in Clark and Marshal (1981), is that people sem to have very litle trouble taking information to be common knowledge betwen themselves and one or more other people. Yet it is strikingly dificult for people to reason overtly about nested beliefs, and thinking explicitly about reciprocal knowledge is very hard indeed. This kind of thinking develops late in childhood: recursive reasoning even just two levels deep is dificult for children under twelve, and yet much younger children (betwen six and eight years old) sem to use definite references felicitously most of the time (Warden 1976). This discrepancy makes it sem as if something other than iterative recursive propositions are involved in the thought proceses behind these common-ground asesments. Furthermore, there is an alternative proposal that alows us to derive reflexive representations of our common ground with others, but does not rely on it. This is what Clark (1994:94) cals the shared basis definition, to wit: p is common ground for members of C if and only if: 1. every member of C has information that basis b holds; 2. b indicates to every member of C that every member of C has information that b holds; 3. b indicates to members of C that p. 21 This means that we can short circuit the reflexive iteration proces altogether as long as we have the basic capacity to understand ourselves as jointly attending to some shared referent, whether it is directly acesible in our shared visual field or some more abstract or remembered entity or event. Thus language serves both as a tool for achieving joint atention, in that it is a means of directing and shaping the atentions of one?s interlocutor to bring them in line with one?s own; and as a highly underspecified system underlying the joint activity of communication, which relies on the support of joint atention to function. On, then, to the nature of joint atention. Atention First, a very brief discussion of atention in general. Wiliam James (1890, reprinted 2007:403) famously observed, ?Every one knows what atention is.? Today it would be more acurate to say that everyone knows that modern studies of atention wil repeat this quote in order to dispute it. Atention is not, it turns out, such a simple thing to pin down, and it?s not clear that there realy is a single psychological?let alone a single neurological?phenomenon of atending to something. Fortunately, the study at hand does not hinge on the more contested aspects of what atention might be. For our purposes, a working definition should suffice: To atend to something is to concentrate on some features of the environment or of a mental representation to the relative exclusion of others. Objects that we atend to in this sense are perceptualy enhanced, and we respond more rapidly to changes to these objects, while our awarenes of other aspects of our visual field is partialy suppresed. This is the 22 concept of selectivity. At any given moment, a person?s awarenes includes only a tiny subset of the many, many stimuli bombarding his or her perceptual systems. In addition to perceptual selectivity, there are other qualities that sem to define atention. For example, there sem to be limits to our atentional capacity; some things that are perfectly easy to do in isolation are dificult or impossible to do simultaneously, even when they aren?t physicaly incompatible. These include thinking, remembering, analyzing new perceptual input, and planning motor activities?in some real sense we have a finite supply of atention that we can ?pay? to these tasks, and when we?ve alocated it to one task or stimulus, there isn?t enough ?left over? for another. Questions therefore arise over whether atention has a single focus: can atention be split into multiple ?spotlights?? If so, how many? How focused is that atention? We also have trouble filtering out certain stimuli?they sem to demand some degre of our procesing capacity, even when we try to pay them no mind. Are interference efects like the Stroop efect, then, a product of unwanted stimuli demanding some of our atention as such, or do we want to say instead that they are simply some kind of efect on our atentional capacity or atentional experience? We could also ask what aspects of phenomenological experience should be considered efects of atention (as opposed to sheer products of higher order conceptual proceses). Atending to an object or event afects how quickly it is perceived and whether it is perceived at al; does it also afect how we perceive it? And so on. These questions and many others are real isues for defining atention and the study of its psychological and neural mechanisms. For our purposes, however, a 23 relatively simple notion of selectivity, that is, giving priority to some information over others, can serve as the defining feature of what it is to atend to something. Further, for the most part (though not entirely) we wil be concerned with intentional selectivity. Selective atention is clasicaly understood to come in two basic varieties: pasive and active. Passive, or spontaneous, atention occurs when some previously unatended-to stimulus captures our atention. Roughly speaking, pasive atention is what happens when we notice something, as opposed to deliberately turning our atention to it. James (1890) observed that stimuli can of course elicit our atention based on their ?imediate? sensory characteristics, such as high intensity, but also by ?derived,? i.e. semantic, features. He ilustrates this distinction by the example of a faint tap; in itself such a sound is not intrinsicaly salient?in many circumstances it wil not stand out against the background mas of stimuli. However, if one is expecting some signal (a lover tapping on the windowpane, for example) even a very quiet sound wil not go un-noticed. As I?ve already mentioned, one of the asumptions we tend to make about other people is that what is perceptible to us wil be perceptible to other people in the same situation. Similarly, we tend to asume that what is noticeable, or salient, to us, wil also catch the atention of another person in the same position. Active attention is a more volitionaly, or at least intentionaly, directed proces. (More on the distinction betwen volitional and intentional in a moment.) In the literature on joint atention as a foundational part of social cognition, atention is generaly (se, e.g. Tomaselo 1995:104) understood as ?intentional perception.? 24 That is, the most important aspect of atention for these claims about social cognition and development is the fact that people atend to diferent aspects of their environment in significantly diferent ways depending on their goals. The proposal is that long before they are able to form explicit understandings of the beliefs of other people, infants are able to understand others as intentional agents in terms of appreciating their imediate, concrete goals and the sensorimotor, atentional behaviors they produce in the course of achieving those goals. Finaly, atentional behavior describes the externaly perceptible actions that acompany the act of atending. Throughout this study, and generaly in research on joint atention, atention wil be understood primarily not as a ?pure? mental state but as a both mental and behavioral activity of atending. Joint Atention What It Isn?t Joint atention is more than mere simultaneous atention to the same thing. For example, events in which some event or feature of the environment is salient to two entities and thus captures their pasive atention do not in themselves qualify as examples of joint atention. If two cats are siting in a room and both se a moth fliter by, they may both be atracted to the moth (as naturaly interesting, tasty prey). However, while they wil then most likely both look at the moth at the same time, they wil have no sense that they are both doing so. Similarly, if they happen by chance to direct their active atention to the same object, there is no joint atention. 25 The cats are not monitoring one another?s atention and have no knowledge of the shared atentional focus. Furthermore, gaze following alone does not qualify as true joint atention. Say that one dog is looking at a juicy piece of meat siting on the kitchen counter. Another dog may observe the first dog?s visual orientation and follow the direction of her gaze because he has learned that when he does so, he often ses something of interest. The atention is stil not joint. Indeed, it is not even mutual; the first dog hasn?t noticed anything about the atentional behavior of the second one. What It Is Joint atention criterialy involves a shared intentional relation to the world. In its simplest form, it is an event in which two (or more) people engage in an interaction that is mediated by some object, while both participants continualy monitor one another?s atention to both the object and to themselves. Further, the participants must mutualy recognize that the atention is shared. To be able to participate in a joint atentional scene, then, one must be able to understand both oneself and the other participant in some way ?from the outside,? as intentional agents. This kind of joint engagement, or triangulation of intentional perception, establishes a joint atentional frame (Tomaselo 2000) within which communication may take place. This frame is defined through the participants? shared understanding of the goal-directed activities in which they are jointly engaged. So, for example, if a child is playing with some blocks, she is also perceiving other things in her environment: the rug she is siting on, her itchy shirt, the window through which sunlight is shining 26 into the room. If an adult comes into the room and joins her in playing with the blocks, the shirt, rug, and window wil not be part of the joint atentional frame. If the adult had come into the room and helped the child remove her itchy shirt, the blocks would not be part of the joint atentional frame, and the shirt would?because the shirt is in this case part of what ?we? are doing. Figure 1-1, adapted from Tomaselo (2003:29), shows how joint atention is implicated in the structure of a linguistic symbol. Any person can use a linguistic symbol to intend (bold lines) that her interlocutor follow her atention (thin lines) to some external entity, aspect of an external entity, or conceptual structure; that is, to share atention to it. In Figure 1-1, the person on the left is refering to a nearby squirel. Figure 1-1: Joint atention and symbolic communication 27 The emergence of this particular kind of joint activity around the end of the first year of life is wel documented (se, for example, Hay 1979; Bruner 1983; Bakeman and Adamson 1984) as is the support it provides for infants? first ventures in linguistic communication. As it is ilustrated here, and in its earliest and most fundamental forms, this interaction involves reference to an object that is directly mutualy perceptible to the participants in the communicative act. For most of us, the foundational experience of joint atention is specificaly and centraly visual. The notion that there is a primary or basic experiential link betwen intersubjectivity and shared seing wil be important later on, in discussion of literary texts that use scenes of simultaneous seing as signals of intersubjective experience. It is nonetheles true that joint atention can also take place in the absence of vision. Congenitaly blind children, for example, develop joint atention in infancy, though its emergence is delayed (Bigelow 2003). The dificulties that blind children have in ataining and especialy in initiating scenes of joint atention confirm that vision is indeed important for and useful in establishing joint atention. But their eventual succeses also afirm that vision is not absolutely required, even as a developmental starting point. And once language enters the picture, both blind and sighted people can abstract away from the basic joint atentional triangle with ease. Language fres us from our imediate shared perceptual seting, alowing us to coordinate atention to distal, remembered, imagined, or abstract objects and events. The emergence of children?s earliest skils of joint atention and intention- reading correlate highly with their earliest skils of language production and comprehension (se reviews in Carpenter, Nagel, and Tomaselo 1998 and 28 Tomaselo 2000). The rich context that comes with joint atention can provide children with the extra information they need to asociate meanings with signals, and to evaluate and compare individual utterances in the course of drawing the necesary generalizations required to acquire a flexible set of abstract constructions and be able to produce and understand novel utterances. Children as young as 12 months, for example, wil spontaneously check where a speaker is looking when she says a word that is new to the child, and link the word with the focus of the speaker?s gaze (Baldwin 1993). Intentions At this stage, we also need a working definition of intentions, intentionality, and a few of the diferent varieties of intentions (cf. Gibbs 1998) that people can have, particularly those types that are particularly implicated in language use and communication. As with atention, intentions and intentionality appear to be intuitively straightforward concepts, yet they admit a great deal of debate and discussion in both philosophy and psychology. Some discussion and explication wil thus be necesary. Searle (1983) defines the concept of intentionality to refer to the quality of directednes or ?aboutnes? that may be a feature of any conscious mental state. In this sense, intentionality obtains for any mental state, or its expresion, that refers to things in the world, or in an imagined version of the world. Intentions to act are particularly obviously and canonicaly directed in this way, but beliefs, wishes, desires, and percepts are also intentional in this sense. Stil, not al mental experiences 29 acesible to introspection are intentional. Fre-floating dispositions, for example, do not qualify as an intentional state. Thus, anticipating a particular event would be an intentional state, but a general sense of anticipation would not. To understand someone as an intentional being, then, in Searle?s sense, is simply to understand them as having mental states that are directed to, or refer to, things in the world, or at least in their conception of the world. Volitional states are therefore intentional states, but not al intentional states are volitional. Intentions, meanwhile, are mental states that are favorably inclined toward bringing about, preventing, or maintaining some state of afairs. Behaviors that are performed acidentaly, then, would be those that are not preceded by a corresponding volition or intention, while a purposeful act is one done with the intention to do so. This distinction is perfectly useful in practical terms, but I should note that it is a source of some philosophical disputes of long standing (se, e.g. Ryle 1949) on a couple of points. The main argument is over an isue of regres, the dilema being as follows: Volitions, or intentions to act, are themselves either voluntary or not. If they are not voluntary, it sems fruitles to say that the actions they cause are voluntary. However, if these intentions are voluntary, it sems that we must postulate for each one a volition to form that volition, and so on, ad infinitum. Another objection postulates that the definition succumbs to a Sorites paradox, due to the fact that many actions sem to happen at a level too automatic to be characterized in terms of intentional states, while stil being voluntary and even intentional in the everyday sense?for example, putting one foot in front of the other while walking. 30 Happily, the question of whether, when, and how children come to understand other people as having intentions depends not a whit on whether the folk notion of ?intention? is philosophicaly sound. For our purposes, this debate can be safely tabled, and we can proced with a working and workable understanding of intention as a mental state in which one means to change (or prevent or sustain) some state of afairs, either in the world or in one?s own mind. More precisely, I wil follow Bratman (1989) and Tomaselo et al. (2005) in defining an intention as a plan of action that is chosen and commited to in pursuit of some goal. Finaly, note also that the use of ?intentional? in the above-cited phrase ?intentional perception? (Tomaselo 1995) sems to be geting at something other than the Searlean sense of intentionality. It also sems to refer to something other than, say, any perception that is the direct result of an intention to perceive. Indeed, in work on joint atention, this phrase is used neither to embrace al perceptions that give rise to a mental state directed towards the world, nor simply to describe a deliberate (i.e. ?active?) atention. Instead, it refers to any atentional behavior taken pursuant to enacting an intended goal. The idea is that someone who intends to climb a tre wil atend to diferent aspects of the tre, in a diferent order, than someone who intends to sit underneath it. Someone who intends to pluck an apple from one of its branches wil atend to it diferently yet. Thus our visual atention is modulated in service of, and serves as an observable correlate to, our plans of action. This goal-oriented atending is what joint atention researchers cal intentional perception. 31 I wil now quickly define two particular sorts of intentions that wil be important later on, namely communicative intentions and joint intentions. The technical notion of communicative intentions comes from the work of Grice (1967) and his atempt to distinguish betwen communication proper and other kinds of transfers of information. His distinction betwen natural meaning, as in ?Those spots mean measles? and non-natural meaning, which is the kind of meaning involved in communicative acts, relies on postulating the existence of a special kind of intention. In Grice?s definition, communication consists of a circumstance in which the ?sender,? or communicator, intends some recipient to think or do something, sheerly by geting her to recognize that the sender is trying to cause that thought or action. In at least two diferent circumstances, then, I may get your atention by clearing my throat, but in only one wil that act of throat-clearing count as communicative. If I incidentaly clear my throat (because it tickles), causing you to notice that I?m in the room, your awarenes of my presence is simply a natural consequence of a non-communicative event. However, if I clear my throat with the intention that you notice me and recognize my throat-clearing as an intentional signal for you to acknowledge me, it is communicative. More simply, ?communication is a complex kind of intention that is achieved or satisfied just by being recognized? (Levinson 1983:16). You may decline to take up the thought or action that a communicative act is intended to produce; but as long as you correctly recognize it as the product of a communicative intention, it has been communicated. If you do not, it has not. Thus, a basic understanding of intentions would sem to be crucial to participating in communication. 32 Joint intentions are intentions that people form in coordination with one another to perform joint actions or to jointly bring about a certain state. There have been some atempts to formaly define joint intentions within the framework of analytic philosophy, the most extensive of which can be found in Tuomela (2007), who ultimately proposes the succinct definition ?The participants jointly se to it as a group that X? (p. 104). Shared intentionality thus refers to collaborative interactions in which the participants form some kind of shared commitment, or goal, and coordinate a plan of action for pursuing that goal. Joint actions can be simple or complex: shaking hands, going for a walk together, dancing a pas de deux, building a skyscraper, having a conversation. Bratman (1992) proposes that there are thre defining characteristics that distinguish these kinds of activities from other kinds of social interaction. First, they require their participants to be mutualy responsive to one another and to provide mutual support?that is, that one wil when necesary take on actions that have as their only purpose the support of another participants? sub- goals. Second, they involve a shared goal; each participant intends and takes it to be common ground betwen the participants that they are to do X together. Finaly, joint actions and joint intentions to undertake them require that the participants must be able to take a step back from both roles in the collaboration. As Tomaselo et al. (2005:681) put it, ?collaborative activities require both an alignment of self with other in order to form the shared goal, and also diferentiation of self with other in order to understand and coordinate the difering but complementary roles in the joint intention.? The ability to share intentions in this way is thus a very special and sophisticated kind of intentional understanding. 33 It is argued, folowing Tomaselo et al. (2005), that this ability is a distinct level of social engagement, crucial for language use, implicit in true joint atention, and unique to humans. Many nonhuman primates understand quite a lot about intentional action; there is some evidence, for example, that apes understand the diference betwen trying to do something and failing, and refusing to do it (Cal et al. 2004). Apes also sem to understand that others se things (Cal et al. 1998) and that what others se may afect what they do (Hare et al. 2001). But there are limits to their understanding of intentions, and similarly there are quite strong limits on the extent to which they share psychological states with others, and most notably, on the degre to which they engage with others in shared endeavors, including basic declarative communication about third entities: no pointing at, showing, or active offering of objects to their conspecifics (Tomaselo and Cal 1997). Developmental Timeline My argument in this study is not only that the dynamics of joint atention are reflected in the structure and content of literary discourse, but also that joint atention constitutes a theoreticaly significant and useful concept for analyzing that discourse. Part of the justification for this claim is the fact that joint atention, especialy joint visual atention, is a central ingredient in our primary experiences of intersubjectivity and reference from long before we are sophisticated enough to reflect self- consciously on those experiences. The experiences of refering, of thinking intersubjectively, and of gazing in concert with another person are connected in a profound and entrenched way from early in childhood development. 34 Infants engage in some forms of social coordination from an extremely early age. By their sixth or seventh wek, infants regularly engage in protoconversations, a wel-documented early form of prelinguistic, temporaly synchronized interaction betwen infants and their caregivers. These interactions involve a back and forth exchange of facial expresions, bodily movements, and vocalizations, produced by both caregiver and child. Even in the first weks of life, these interactions show systematic paterns (Bateson 1975; Trevarthen 1979; Bruner 1983) of turn-taking exchanges and an addres-and-reply structure. The ability to fully participate in joint atention, however, depends on a suite of prerequisite skils on top of basic coordination with another person. These include the ability to detect and track the atentional behaviors of other people; the ability to direct or manipulate the atentional behavior of other people; and the ability to understand themselves as intentional agents, and other people as intentional agents like themselves. These skils emerge over the course of the first two years of life. Tomaselo argues that a crucial transition occurs around the end of children?s first year of life, in which several linked behavioral changes appear in relatively tight synchrony, in the window betwen nine and twelve months of age (Tomaselo 1995, 2003; Tomaselo et al 2005). This stage in development, in which complex social and symbolic skils of imitative learning, declarative gestures, and social referencing (i.e., the use of another' person?s perception of some state of afairs for developing one?s own understanding of that situation) appear nearly simultaneously, is argued to mark a qualitative change in the nature of children?s behavior, reflecting a newfound ability to take the intentional stance. To say that a child has this ability does not require us 35 to make any strong claims about the nature of her mental representations of other people?there is no implied commitment to a particular position on whether, for example, she has any explicit theory of mind. Rather, folowing Dennet (1987), we can frame this claim somewhat more conservatively, to say that these young children evidence a predictive strategy of interpretation that presupposes that other people act intentionaly. Milestone developments in the development of the intentional stance include the progresion from learning about natural afordances of objects to learning their intentional afordances, and eventualy decoupling the two (as sen very clearly in pretend play where, for example, a child can ?hamer? with a carot). Children miic adults? behavior beginning in early infancy, but it is only around the end of the ninth month, generaly, that they begin to reproduce an adult?s goal-directed actions on outside objects (Carpenter, Nagel and Tomaselo 1998). This is a major leap forward. Studies of children?s imitative behavior which control for the diference betwen genuine imitation of intentional behavior and simply reproducing an adult?s efect on some object suggest a two-stage developmental path. In emulative learning, i.e. simple miickry, infants can se an adult manipulating some object and perhaps learn something about the physical afordances of that object that they didn?t know: for instance, that a box can tip over. In imitative learning, interactions with objects are understood in light of larger intentional relationships: an understanding of what an object is used ?for?. Infants become true imitators soon after their first birthday, by around fourten months of age. Carpenter, Akhtar, and Tomaselo (1998), for 36 example, found that while children under eleven months old would reproduce an adult?s novel, unusual action as long as it produced an interesting result, sometime betwen eleven and fourten months the majority of infants began both to reproduce the action and to visualy check for the interesting result in anticipation, suggesting a new understanding of behavior as goal-directed action. They also found that fourten- to eighten-month-olds distinguished betwen actions that semed to be mistakes and those that semed purposeful, reproducing ?intentional? sequences about twice as often as ?acidental? ones. Meltzoff (1995) found that by their eightenth month, infants presented with demonstrations in which an adult tried but failed to achieve the end results of some target actions (say, trying but failing to pull two halves of an object apart) reproduced the target actions through to their conclusion, even though they had never sen the full target sequence, just as wel as a control group who saw the adults perform it succesfully. Around this same age, infants also begin to engage in significant amounts of coordinated joint engagement (Bakeman and Adamson 1984), activities in which the child and an adult interact together with some object, and the child takes an active role in the interaction. Nine-month-olds often partake in ?pasive joint? activities? for example, taking an interest in a ratle that a caregiver is shaking for them?in which child and adult are actively involved with a single object at the same time, but the child evidences litle interest in the adult?s involvement in the scene. This is a notable contrast to the kind of shared engagement found at twelve to fiften months, when children coordinate with the adult, directing adults? behavior and responding to their direction. Ross and Lollis (1987) also found that infants of this age responded to 37 an adult who paused in a shared activity by prompting her to re-engage, and sometimes by taking the adult?s turn for her, suggesting that they have some sense of coordinating diferent roles in the pursuit of a single shared goal. In these interactions, children coordinate not just actions but perceptions with their partners; coordinating intentions and coordinating atention are closely tied together. As with early social coordination, infants distinguish and react to others? visual atention when it is directed at them wel before the magic age of nine to twelve months. Babies in their first few months of life already show a strong preference for faces and face-like arangements, and evidence the ability to recognize and sustain eye contact. They generaly engage in protracted sesions of mutual gazing with adults by the end of their third month, and from at least two months of age, infants smile more when adults are looking at them than when they are looking away, and the onset of mutual gaze is a reliable and powerful elicitor of infant smiles (Reddy 2000). It takes a few months more before infants begin to engage in gaze following. Six-month-old infants sem perhaps to do some very large-grained gaze following, looking to one side of the room or another depending on the orientation of an adult?s head (Butterworth and Jaret 1991), though Corkum and Moore (1995) find that before about 10 months of age, infants were as likely to respond to adult head turns by looking in the opposite direction. The granularity and acuracy of gaze following improve over the next several months. These improvements are also acompanied by increasing interpersonal sophistication. Carpenter, Nagel, and Tomaselo (1998) identified a thre-stage presentation of joint atentional interactions, clustered around childrens? first 38 birthday, emerging in a very consistent order. Starting around nine months of age, children reliably check the direction of an adults? gaze. Following their mastery of this task, usualy by the age of twelve months, children can reliably pick out the object an adult is looking at, as long as it is also within their field of vision, and their acuracy increases if the adult also points toward the object. Finaly, in the first few months of their second year, they begin intentionaly to direct the atention of others. Early language production and related pre-linguistic declarative behaviors are linked to developing mastery of the joint atentional frame. New communicative skils arive in a sequence demonstrating increasingly sophisticated manipulation of and direction of other people?s atention. Acts of imperative pointing, when children point at an object when they want an adult to do something for them, first appear at the age of nine months (Baron-Cohen 1997), when infants are just starting to check the general direction of an adult?s gaze. Much as nine-month-olds? gaze following is just on the cusp of significance, these first pointing gestures are ambiguously communicative at best. In their earliest uses, they sem not to reflect a desire to produce an efect in another person, but to serve as a way for the infant to orient herself toward an object she desires or that interests her. Some infants point when no one is in the room with them, for example; Franco and Butterworth (1996) find, too, that when many infants first begin to point, they do not monitor the reaction of adults who are in the room with them. But as infants get more sophisticated in their gaze following, they also grow more sophisticated at incorporating it into their atempts at pointing communicatively. The infant begins by pointing and then checking the reaction of a nearby adult; some 39 months after that, in the first half of a child?s second year, they begin to look at the adult first, to se if they have her atention, before pointing. Around the same time that they make this transition, children branch out from purely imperative pointing to making protodeclarative gestures. Protodeclaratives involve pointing at an object or showing it to an adult purely in order to draw the adult?s atention to that object. These gestures typicaly appear about a month before the appearance of declarative words, usualy around twelve months of age. 4 As we have sen, adults rely on the dynamics of joint atention to support the inferences about common ground that underlie their construction and interpretation of utterances. In much the same way, children?s ability to monitor the atention of others, in which they combine the overt indices of visual atention described above with an understanding of various additional dimensions of the communicative scene in which the atending takes place, alows them to make subtle and complex inferences about the referents of new ords. A child learns new ords not just when adults follow her own established focus of atention and explicitly name the object she is atending to, but in the course of al sorts of everyday activities. Her ability to monitor the atention of others, combining the overt indices of visual atention 4 There has been some disagrement over whether twelve-month-olds do in fact engage in genuinely declarative pointing. Franco and Butterworth?s findings about gaze checking suggest that pointing at this stage constitutes a genuinely communicative act, intended to afect the adult in some way. Some skeptics (e.g. Moore and D?Entremont 2001) have argued that the child?s intended efect in these cases is not declarative, but simply to draw atention to herself. However, studies such as Liszkowski et al. 2004, in which twelve-month-olds responded more positively and pointed more across trials when adults shared atention and interest by alternating gaze betwen child and object than when adults emoted while orienting only toward the child or only toward the object, suggests that these gestures realy are atempts to share atention to the pointed-at object. In any case, they are usualy followed very closely in development by the appearance of the declarative use of individual words. 40 described above with an understanding of various additional dimensions of the communicative scene in which the atending takes place, alows her to make subtle and complex inferences about the referents of new ords. Learning words is an activity that requires another person. Sometimes this happens in the context of a ?naming game? in which an adult points out an object by holding it up or pointing to it, ensuring that the child is atending to it, then teling the child its name. But while this is a stereotypical scenario of new-word learning, it is not, in fact, the primary scenario in which most of a child?s new ords are learned. Such activities are far les common in other cultures than it is in the homes of middle- clas Westerners, and even in setings where they are common, they make up a tiny portion of the opportunities children have, and use, for learning words. For one thing, words denoting things other than the labels of physical objects are not usualy involved in these kinds of point-and-name games. Verbs, for example, occur far more frequently in contexts of directing or anticipating behavior, such as ?Eat your squash,? uttered while directing a spoon toward the child?s mouth, or ?Give me the bal,? uttered with a hand held out to receive it (Tomaselo 1995). While there is asuredly rich context providing clues to the meaning of ?eat? and ?give? in these scenes, making sense of the relationship betwen these words and their referents clearly requires a nuanced and flexible understanding of intention and atention. Studies by Dare Baldwin (1991, 1993), for instance, demonstrated that by the time they are at least nineten to twenty months old, children appreciate the significance of line-of-regard in learning new ords. When presented with a situation in which an adult looked at and named an object that the child was not looking at, 41 children of this age spontaneously looked up at the adult to check the direction of her gaze, and succesfully learned the new ord as the name for the object of the speaker?s focus, rather than the original focus of their own atention. Infants of the same age performed equaly wel in the face of other competing cues for referentiality and learned new object labels correctly simply by noting which of several available objects an adult was looking at when she said the new ord, even when the object was hidden, for example by looking into a bucket while saying the word (Baldwin 1993b). Another study (Tomaselo and Akhtar 1995) demonstrated the subtlety with which young children are able to combine gaze-monitoring with other social- pragmatic clues to diferentiate whether a new ord refers to an object or an action. In this study, an adult and a child began the experiment by playing a game on a mery-go-round and then moving on to some other activity. Later, they returned to the mery-go-round. In one condition, the adult first readied the mery-go-round for play and then held out an object to the child. While alternating her gaze betwen the child and the mery-go-round, the adult said, ?Widget, Jason, widget.? In the other condition, the adult held out the object to the child and alternated her gaze betwen the child and the object (not looking at the mery-go-round). In the first condition, the children treated widget as a request to use the toy with the mery-go-round, learning the word as the name for the action asociated with the mery-go-round structure. In the second, the children learned widget as the name of the object, not the action. Taken together, these findings suggest that joint atention and its underlying skils of social cognition are a crucial precursor to expresing and interpreting 42 communicative intentions. What?s more, they ilustrate that our foundational experiences of refering, of communicating, and of understanding other people as intentional, thinking agents like ourselves are inextricably bound up with the experience of jointly atending. For most people, this means that there is a strong and early experiential correlation among thre things: ?syncing up? our beliefs and intentions with another person, performing symbolic reference (that is, semiosis), and the act of simultaneous looking at some external object. As we proced, we wil find that literature, particularly modernist literature, taps directly into this close-knit asociation. 43 Chapter 3: Participants in the Literary Event From Conversation to Literature The capacity for joint atention lies at the very foundations of linguistic communication. The ability to think intersubjectively, in tandem with other people and about the outside world, is at the heart of not just the acquisition of language and understanding references to our imediate environment, but also our ability to expres and understand imagined, remembered, or otherwise distant ideas. I claim, further, that literature capitalizes significantly on the fact that communication and understanding other people relies in some fundamental way on triangulating atention to concrete, perceptible objects. But concrete, mutualy perceptible objects would sem to be noticeably absent from the literary event itself. In a situation where readers and writers are unknown to one another, can they realy be said to interact? How do the dynamics of imediate, direct interaction bear on the experience of producing and consuming literary texts? The phenomenon of joint atention gives us a new ay to think about some old questions about the relationship betwen authors and readers?and perhaps some of the other participants in literary discourse as wel. Who counts as a participant in this discourse? Is it appropriate to treat literature as the product of interactions betwen these participants? What kinds of roles do these participants inhabit, and how do those roles vary across times and cultures? How, if at al, do their considerations of one another?or their ideas of one another?afect the way they 44 compose and interpret literary texts? How much room for variation is there in these configurations? Conversation analysis, gesture studies, linguistic anthropology and other investigations of the interactive dimensions of language have iluminated the degre to which meaning and linguistic structure in conversation are inextricable from the situated, interactive occasion of their production. In the words of Goodwin (1979: 98), sentences ?emerge as the products of a proces of interaction betwen speaker and hearer that they mutualy construct the turn at talk.? This valuable insight into the role of collaboration in the construction of meaning runs into some trouble, however, when applied to the consumption of writen, published texts. Writing and reading can be, and often are, understood as a sort of conversation betwen author and reader. However, the very insights into the role of interactivity in language that might make this model useful for theorists also render it problematic. Readers, Authors, and ?The Intentional Fallacy? In the middle of the twentieth century, the nature of the relationship betwen those who produce certain kinds of linguistic output?authors of literature?and those who interpret or understand those utterances?readers?became a central topic of discussion in Anglo-American literary and philosophical circles. Influential work by W.K. Wimsat and Monroe Beardsley (1942, 1954) argued that the intentions of an author were imaterial to the task of the reader, and particularly to the tasks of a special kind of reader: the critic. They coined the term ?intentional falacy? to dismis the premise that critics' readings of literary texts should depend on any external 45 evidence about what an author ?meant? when writing them. This anti-intentionalist stance became strongly asociated with the later period of New Criticism, during its period of dominance in the American academy after the Second World War. This approach to criticism had not only intelectual merits but also practical social and economic ones. Many (e.g. Graf 1987) have pointed out that institutional presures of the period during which New Criticism flourished, especialy in the United States, made particularly fertile soil for their approach to literary scholarship. The G. I. Bil and the expansion of the university system brought an influx of undergraduates who generaly lacked the elite secondary educations of their predecesors. They also had diferent objectives for their university studies and fewer resources for leisurely scholastic pursuits. The notion that literary texts could, and indeed should be studied without demanding extensive historical and biographical knowledge, while preserving academic rigor, was an excelent match for the new demographics of post-secondary education in the post-war period. The anti-intentionalism of ?The Intentional Falacy? was primarily an argument about what kind of evidence it was proper to use in academic and profesional literary criticism, rather than a grander claim about the nature of meaning or linguistic reference. In Wimsat and Beardsley?s view (1954, p. 10-11), critics should base their readings only on the evidence of the text itself, understood by reference to ?public? linguistic and cultural information, ?our habitual knowledge of the language, through gramars, dictionaries, and al the literature which is the source of dictionaries, in general through al that makes a language and culture.? The kind of information that was verboten was the ?private and idiosyncratic,? such as ? 46 revelations (in journals, for example, or leters or reported conversations) about how or why the poet wrote the poem?to what lady, while siting on what lawn, or at the death of what friend or brother.? In a hazy grey area came the kind of information they termed ?semiprivate,? regarding such things as information about idiosyncratic meanings of words within an author?s coterie. One might justifiably draw on this kind of biographical knowledge in crafting a proper reading, as long as one constrained oneself to what Wimsat and Beardsley deemed ?linguistic facts,? taking care to avoid using biography as evidence of ?what the author intended,? rather than ?the meaning of his words and the dramatic character of his utterance.? In its original form, the anti-intentionalist stance was mainly an argument about what one ought to do, not about what a text is, what meaning is, or what readers and authors realy are, nor a descriptive claim about what readers or authors actualy or normaly do. It was also a claim primarily about a particular kind of reading, rather than reading in general: a template for behavior for people who would write or talk about texts in generating academic criticism, not for one?s personal, private comprehension, appreciation, and pleasure. Furthermore, the sort of academic criticism Wimsat and Beardsley were addresing involved a diferent set of functions than are usual for the field today. Avoiding the intentional falacy was not a program for how one ought to interpret a text, but specificaly how one ought to interpret it for the sake of judging its artistic merits, these days a project consigned to reviewers more than scholars. Stil, while ?The Intentional Falacy? was presented more as a statement of profesional standards than as a theory of language, it necesarily relies on a model in which intention and meaning are separable. 47 The notion of the intentional falacy is stil alive and wel in critical theory, though its import has changed somewhat, acompanying the decline of some of the context for Wimsat and Beardsley?s proposal. This is not to say that contemporary criticism is uniformly opposed to discussions of authorial intention. Steven Knapp and Walter Benn Michaels, in their articles ?Against Theory? (1982) and ?Against Theory 2? (1987), for example, take a radicaly intentionalist approach, claiming that a text means exactly and only what its author intends it to mean. Knapp and Michaels undertake this argument largely in support of a larger claim that the general project of literary theory is both misguided and a distraction from critics? proper job of interpreting texts: Theory ?is the name for al the ways people have tried to stand outside practice in order to govern practice from without. Our thesis has been that no one can reach a position outside practice, that theorists should stop trying, and that the theoretical enterprise should therefore come to an end? (1982:742). While few are wiling to adopt quite such a radicaly intentionalist and radicaly anti-theoretical position as this, critics and theorists of al stripes these days do tend to take at least a middle-ground approach to questions of authorial intention: certainly information about an author?s disposition, larger projects, and other clues about an author?s intentions are widely considered fair game for critical atention. To be interested in the political, cultural, and ethical dimensions of literature, as cultural studies and postcolonial approaches urge us to do, requires critics to acknowledge and contemplate at least some aspects of intentionality. This line of critical interpretation takes the role of intention very seriously, but also tends to consider it only at certain levels of specificity. Theoretical questions regarding the 48 nature, importance, and consequences of diferent kinds of intentions are on the table in these critical traditions. The specific intentions of individuals may also come under discussion, but these are mainly of interest at the level of authors? general dispositions towards certain subjects, or regarding the nature of the larger project that a text was intended to fulfil. These often appear in naratives of the conflicted, hidden, shifting, and politicaly interesting underlying intentions that can be diagnosed from or taken to be informative for the interpretation of a given text. Thus, in this mode, one might discuss, with Edward Said (1975:153-158), the ways in which Seven Pilars of Wisdom reflects the divided purpose of its author, T. E. Lawrence, noting various features of the text can be explained via an analysis of Lawrence?s conflicting goals, in which an impulse toward confesion perpetualy overides the original or base project of a straightforward, impersonal history. Communicative intentions are not considered determinative of interpretations in this tradition, but their diagnosis, at least, can provide rich material for literary scholarship. At the same time, the notion of the intentional falacy has had a lasting influence on the kinds of arguments that critics make and the way that they make them, not to mention the way that we teach our students to talk about the texts that they read in our clasrooms. In some approaches, the disavowal of authorial intention is more radical. In general, practitioners of cultural theory, literary criticism, and intelectual history who work in a ?postfoundational? mode?that is, influenced by poststructuralist, postcolonial, and postmodern epistemology?are wary of ascribing any privileged role to authorial intention in discussions of texts and their meanings (se, e.g., the May 2002 special isue of History and Theory, ?Forum on 49 Intentionalism in Intelectual History?). For deconstruction, texts necesarily elude any authorial control: there are no privileged points?neither authorial intention, convention, nor context?that determine the significance of a text, because the appeal to coherent meaning itself is a falacy. Critics often take pains to sidestep the intentional falacy by employing various practices of intelectual hygiene that alow the critic to avoid making presumably insupportable claims about the actual thoughts and intentions of an actual author. The ?implied author? suggested by Wayne Booth (1961) is one such useful construct: not the actual flesh-and-blood person who wrote the work in question, but a hypothetical ?second self? evoked by the text itself, to whom critics can frely atribute any ?moral and emotional content? (p. 73) of a work that does not sem to originate with its narator, without recourse to problematic claims about the author?s own thoughts. Nor is this kind of atempt to insulate interpretation from appeal to authorial intention limited to the realm of academic criticism and aesthetics. The interpretation of legal statutes, for example, is a domain of textual analysis that typicaly has a great deal more imediate practical efect than literary interpretation, and its history is intimately tied up with questions of authorial intention. 5 For most of the twentieth century, judges in the United States typicaly examined both a statute?s text and its legislative history in an atempt to determine the original intent of the enacting legislature. When confronted with circumstances that the drafters clearly did not or could not forese, they atempted to approximate the general purpose or spirit of the 5 Se DeSanctis (2004) for an extended discussion of the feasibility of New Textualist interpretation given findings in linguistic pragmatics regarding the role of intention in meaning construction. 50 statute. Indeed, a long line of Supreme Court decisions established a precedent that the ?sole task? of the Court in such maters was to determine congresional intent or purpose (Eskridge 1990). More recently, however, the U.S. judiciary has sen a movement away from intentionalism and toward a particularly stringent version of clasic textualism, in which the judge?s goal is to apply a statue?s ?plain meaning? (Eskridge and Frickey 1995:514). This ?New Textualism? is identified primarily with United States Supreme Court Justice Antonin Scalia, Chief Justice Wiliam Rehnquist, and United States Court of Appeals Judge Frank Easterbrook, and rests on the premise that the law consists only of the ratified text of a statute. That text should be interpreted on the basis of no external sources beyond case precedent, established textual canons (a handful of rules of thumb of long standing for determining the meaning of words in a text), and ?ordinary? use of the language, commonly determined by reference to dictionary definitions (Eskridge 1990). Like Wimsat and Beardsley, the New Textualists consider extra-textual evidence of the authors? intentions (such as evidence of how the statute has been previously construed by the drafting agency) completely out of bounds. The presumption is that it is not only desirable but also possible to interpret statutory language independently from al consideration of authorial intent. Also like the New Critics, practitioners of New Textualism rely implicitly on a model of language use in which intention and meaning are detachable and discrete. 51 Intention and Meaning As is no doubt clear by now, I do not concur with this model. A text does mean both more and les than its author intends, but this is just to say that there is more than one kind of meaning and more than one source of it. In linguistics, it is generaly acepted that some elements of meaning are conventional, while other elements are context-dependent. It is perfectly correct to observe that whatever Humpty Dumpty may say he means by ?glory,? ?a nice knock-down argument? is not what it properly, or conventionaly, means. Conventions do not depend on the individual intentions of a particular speaker; instead, they are emergent functions of a larger community: regularities in behavior that are at least partly arbitrary (other linguistic communities could and do have diferent conventions to deal with the same thing) and are common ground for the members of that community. Speakers of English or a given dialect of English know we use the word ?dog? to talk about dogs (Lewis 1969; Clark 1996). And yet when we are confronted with a speaker who sems to intend something other than what convention would suggest her utterance should mean, we stil generaly take an interest in figuring out what she was ?trying to say? as wel as what the conventional entailments of the uttered string of linguistic material might be. Furthermore, believing that language use relies crucialy on understanding communicative intentions doesn?t mean that we have to repudiate the notion that utterances can have significance that their utterer does not intend as wel as the significance that she hopes to convey. Uterances can be symptoms, in other words, as wel as communications. Grice (1957) draws a distinction betwen these two types of 52 significance by proposing two categories of meaning: natural meaning, as in These spots mean measles and non-natural meaning or meaning-n, in which a speaker (S) means-nn something (z) under the following circumstances: S meant-nn z by uttering U if and only if: (i) S intended U to cause some efect z in recipient H (i) S intended (i) to be achieved simply by H recognizing that intention (i) 6 The idea here is that the communicative part of utterance meaning?which is just one component of the broader category of meaning, or significance?consists of a particular kind of intentional action, which is satisfied simply by being recognized. There are several reasons to believe that there is no way for people to understand texts without some recourse to thinking about the communicative intentions that motivated their production. As Gibbs (1999: 16) put it, Our interest in communing with the intentions of others is so deeply a part of how people construct meaningful interpretations? that we sometimes fel that the search for intentions is optional and therefore can be abandoned if desired. Yet an explicit search for the psychological underpinnings of human action wil reveal the fundamental importance of communicative intentions in many aspects of meaningful experience. In order to make sense of much of what goes on in a conversation, one must think of one?s interlocutor as someone with intentions much like one?s own. For example, when I say ?It?s hot in here,? my addrese may wel be able to succesfully recover my communicative intentions on multiple diferent levels, using her knowledge of the 6 This is a slight reformulation of Grice?s definition, taken from Levinson (1983: 16). 53 conventions of English, of the social conventions of the various communities to which we both belong, our surroundings and circumstances, of what I am likely to reasonably asume about her own thoughts, beliefs, and atention, and so on. She is likely to understand that I made these noises deliberately, with the intention that she recognize them as English and directed to her. She may also correctly understand that I intend for her to adopt my belief that the room is warmer than is comfortable, and that I further intend to induce her to open a window. It is also of course possible to make mistakes about the intentions of one?s interlocutor. Sometimes these misunderstandings can be repaired. In the example above, for example, my addrese may take me to mean that I would like her to open a window, while I meant only to make an observation. Her response to my remark wil tel me something about the way she construed it. Suppose she says, ?I?m so sorry!? and rushes to the window. I can correct her about my intentions: ?Oh, I didn?t mean you needed to do anything.? Or I can change my mind about what my utterance should be taken to mean, and just say ?Thank you.? Either way, we are both making some efort to discern the intentions behind the other person?s utterances, and both expecting to be treated as if our communicative intentions were an important part of our contributions to the conversation. Theoretical Commitments: Language Use is a Joint Activity The General Idea If a model of language use that separates intention and meaning is not satisfactory, we wil need something else. The model I endorse is one based on 54 research in pragmatics, sociolinguistics, cognitive psychology, and philosophy of language, especialy the work of Clark (1996) and Lewis (1969). The central claim of this theory is that language use is an activity that is performed jointly by interlocutors. Joint actions are performed by ensembles of two or more people in coordination with one another. Balroom dancing, playing a duet, shaking hands, and running a race are al joint activities. The agents participating in a joint activity are al doing something more than performing their individual actions. Each one is acting not autonomously, but in concert with the other participant or participants. Clark (1996: 19) cals these individual actions participatory actions, as opposed to the autonomous actions that one might take on one?s own. This is the diference betwen, for example, playing a sonata on the violin, and playing as part of a string quartet. The later performance is not best described as four individuals each playing a piece of music, but as a quartet playing a concerto together. People engage in joint actions in order to cary out joint projects. Shaking hands, for example, is a smal joint project. Jane may extend her hand to Michael while she has his atention, marking the entry point of the joint project of shaking hands. As long as Michael has the necesary background knowledge to understand that she is initiating a handshake, Michael has any number of options for how to respond. For example, he can simply comply, taking up her proposal by shaking hands with her. He can respond with an alteration, say, by extending his left hand instead of his right, or grasping her hand and pulling her in for a hug. He can decline her proposal, perhaps by saying something or by showing her that his hands are full. Or he can withdraw, perhaps by giving her the cut direct. 55 Linguistic symbols are inherently collaborative in that learning to use them requires role-reversal imitation, using them towards others the way others have used them towards you (se, e.g. Tomaselo et al 2004). Language use is also inherently collaborative in the sense that it involves establishing a joint goal betwen interlocutors: to reorient (at least some of) the listener?s intentions and atentions so that they align with the relevant intentions and atentions of the speaker. The communicative function of language is not caried out unles both?or al?participants in a conversation act jointly to bring it of. To the degre that they are communicating with one another, interlocutors, even those who on other levels disagre fervently, collaborate in pursuit of this goal. They do this by, for example, expresing their communicative intentions in ways that that they believe their interlocutor wil understand, and clarifying their intentions when necesary. Similarly, they try to make appropriate and relevant inferences based on what they know about the speaker?s background and knowledge. But speakers and hearers cannot read one another?s minds, which means that setling on mutualy aceptable shared construals is a coordination problem that they must try to solve. Any time that two or more people have interests or goals in common, and the actions they must take to achieve those goals are interdependent, they have a coordination problem. Thomas Scheling introduced this type of problem to game theory in his 1960 study, The Strategy of Conflict. An example of a very simple coordination problem would be a game in which two players are each given $100 and told that they must divide it into two piles, without communicating with one another. If both players divide the money in the same way, both get to keep the hundred 56 dollars; if the two sets of piles difer, neither gets anything. Another would be a situation in which you are meting someone on a given day, but have made no arangements about what time you?re meting and have no way of communicating with him beforehand. Scheling devised a number of such games, and found that people tended very strongly to setle on the same responses. 88 percent of people playing the money game, for example, put the money into equal piles of $50 each. ?Virtualy al? players, in the absence of any other compeling information, said they would go to the meting place at noon. These situations require something other than a simple prediction of what one?s partner wil do. After al, the partner is also trying to figure out what the first person wil do, and so on, apparently endlesly. Instead both partners have to analyze the situation that they have in common and identify some kind of mutualy recognizable clue that they can use to coordinate their expectations. For example, if the players are presented with thre red blocks and one blue block and told to choose just one, the blue one wil probably strike both of them as the most salient and noteworthy block. It is a good bet that the other person wil also consider the blue block the most noteworthy, and so choosing it is a (likely) winning strategy. Scheling caled this kind of clue, or coordination device, a ?focal point? and observed that succesfully co-identifying a solution frequently relies ?more on imagination than on logic; it may depend on analogy, precedent, acidental arangement, symmetry, aesthetic or geometric configuration, casuistic reasoning, and who the parties are and what they know about each other? (57). 57 Lewis (1969) had the insight that conventions, including linguistic conventions, are a community?s solution to recurrent coordination problems. Say that you and I frequently want to met for lunch. We might coordinate our plans by gueswork, relying on the fact that there is only one cafeteria in the building where we both work, and that noon is the most noteworthy time of day, not to mention the canonical time for lunch. Or we might coordinate by explicit agrement, agreing to met at the caf? across the stret at 12:30. If we make the same arangements every time we have lunch together, after a while we may stop being explicit about them. We can simply agre to have lunch and go to the caf? at 12:30, confident in our mutual expectation of one another?s behavior. We have evolved a convention betwen the two of us. Larger communities and whole societies also have evolved conventions to deal with commonly occurring coordination problems. Clark (1996: 71) summarizes and adapts Lewis? definition of a convention as follows (with slight modifications of phrasing). A convention has five properties: 1. It is a regularity in behavior 2. That is partly arbitrary 3. And common ground in a given community 4. As a coordination device 5. For some recurrent coordination problem. For example, the widespread use of English word dog is a convention that is used as a coordination device for talking about dogs. This is a regularity in behavior among English speakers?using dog to speak of dogs is common, widespread, and reliable. This regularity is common ground for the members of the English speaking community. It is arbitrary in that chien, for example, works just as wel for speakers 58 of French. (It is also partly non-arbitrary?for example, dog uses English phonology while chien uses French phonology.) Lewis and Clark both argue that language systems?the phonological, lexical, syntactic, and semantic rules of a language?are a special kind of conventional basis for joint actions, a tour de force example of what Lewis cals signaling systems. Signaling systems give rationales for a particular kind of joint action, in which a speaker produces a signal based on the state of afairs she wants to communicate, and the hearer takes the signal to mean something, based on the conventions of the signaling system. At the same time, the conventions of a language alone are not enough to make communication work. Conventions specify some of the potential uses of a linguistic construction, but the conventional content of a linguistic expresion generaly underspecifies its construal in context. Every example of language in actual use raises some non-conventional coordination problems. Many references are indexical: their referents cannot be resolved without some knowledge of the discourse participants? current common ground, as when I say ?That one was George.? Which one? Which George? We must rely on non-conventional coordination devices to decide. Interlocutors must also resolve ambiguities?expresions typicaly have more than one conventional meaning, and speakers often go beyond even these conventional meanings to generate nonce constructions that depend on context and background for their interpretation. These may rely on background knowledge that is broadly common ground in the culture, such as ?She?s a regular Houdini,? or something far more idiosyncratic. Clark and Clark (1979) provide an amusing 59 example of the later: if Ann tels Bil, ?Max went too far this time and teapotted a policeman,? she is relying on the fact that she and Bil share knowledge of Max?s odd habit of sneaking up on people and rubbing them with teapots. In person, people get imediate fedback about whether they have succeded or failed in their communicative intentions. There are also a host of visual cues that help us to keep track of the imediate atentions of our interlocutors, and this knowledge provides important information about what is appropriate to consider part of the common ground underlying a conversation. Application to Textual Discourse Models of language use that emphasize this interactional factor tend to treat texts as communicative acts that are very much like conversation, in which authors play the part of speakers and readers play the part of hearers. In this acount, authors and readers have to do some extra imaginative work that ordinary conversations may not require, but they are stil in some fundamental sense interacting with one another. For example, Clark (1996) suggests that the authors and readers of fictional texts ?jointly pretend? that the communicative acts depicted by a text are realy taking place. Bruce (1981) writes of a ?social interaction betwen author and reader? in which readers understand authors to be depicting a communicative interaction betwen an implied author and an implied reader, which itself depicts an interaction betwen a narator and a narate. There are some obvious limitations to this analogy. For example, the production and comprehension of literary texts lack most of the concrete 60 opportunities for direct interaction that are available in face-to-face communication. The readers of published works generaly have no way of confirming what they have understood or asking for clarification, and authors have no way of modifying their contributions on the fly in response to the reactions of their readers. At the same time, there are many excelent reasons to believe that the same kinds of conceptual work that underlie language use in conversation are crucialy involved in the production and interpretation of novels, screnplays, esays, and any number of other uses of language in ?nonbasic? setings (Clark and Brennan 1991). An enormous amount of work in cognitive linguistics has argued convincingly that the language of literature and the language of everyday conversation are expresions of the same cognitive mechanisms. Al of these cases involve human beings with the same social and cognitive capacities and limitations, using a hugely overlapping set of linguistic resources: lexical, semantic, gramatical, and pragmatic. A published story or advertisement can use indexical expresions, generate implicatures, and trigger presuppositions. It can do rhetorical work and perform a variety of speech acts: it can exhort, request, perhaps even promise. Furthermore, it is excedingly common for authors and readers to act as if reading a novel has a great deal in common with being on the hearing end of a conversation, and as if writing one has a great deal in common with speaking to someone. Readers use what they know?or think they know?about an author in generating inferences and making judgments about what a text ?means? (se, for example, Nolen 1995 and Gibbs 1999). Similarly, readers often draw or atempt to draw explicit inferences about authorial intent as they read: does the author mean for 61 this to be ironic? Is this supposed to be funny? There is clearly some significant relationship betwen meanings that are generated in face-to-face conversation and those asociated with literary texts; the question, of course, is what exactly this relationship might be. Theoretical Commitments I: On the Nature of ?Texts? and ?Discourses? Text I have been acting so far as if the main salient diference betwen the sort of language use exemplified by face-to-face conversations and the sort that is exemplified by the novel is that the later is writen. The primary distinguishing feature of this kind of discourse might sem to be that it has authors and readers, rather than speakers and hearers. The questions to be answered appear to be: Is reading realy like being on the receiving end of a conversation? Is writing realy like talking to someone? This is not an uncommon way of defining the problem. Dixon and Bortolussi (2001: 1), for example, criticize the application of conversational models to text procesing by arguing that there are crucial ways in which ?many forms of writen discourse? are ?unlike oral comunication? and like one another. Chafe (1994: 224) writes of ?the desituatednes of writing? and observes that ?[t]he writing situation is itself unreal in its detachment from the co-presence and interaction which are normal for conversational language.? However, I want to pause here to point out that while many of the discourse types at isue here are indeed writen rather than spoken, the sheer fact that they are 62 writing is not the most important thing they have in common. Instead, I want to draw atention to the fact that texts are broadcast, or published. The questions above have unstated asumptions. They might be more explicitly phrased: Is reading published texts realy like being on the receiving end of a conversation? Is writing for publication realy like talking to someone? I believe that this is the more important distinction to draw for the theoretical questions at hand. It is worth noting that people can be unambiguously engaged in direct, conversational interaction in writen media. For example, my neighbor and I can scribble notes to one another during a colloquium. When we do this, we are clearly addresing our writen uterances to one another. We can also se one another, and share a visible and audible, jointly acesible shared environment. We can imediately provide both linguistic and non-linguistic indications of whether we are atending to and understanding what the other person is writing. Conversations conducted via instant mesaging, another writen arena of conversational language use, usualy lack some of these features: typicaly the participants are not visible to one another and do not share a common perceptual ground. They are, however, known to one another in the same way that any ordinary participants in a conversation would be. They may be strangers, but they understand themselves and one another to be directing their utterances to a specific individual or individuals. They can also respond to one another imediately. They can interupt, ask for clarification, take up or reject proposed construals, and modify what they say quickly and flexibly in response to one another?s contributions. 63 Private leters lack the conversational qualities of co-presence and simultaneity. Nonetheles, if al goes wel, and the leter is delivered and read as intended, I think most analysts would agre that the writer and addrese are engaged in a real, if distant, interaction with one another. Anyone who intercepted the leter before it reached its intended destination, or who happened across the leter after the fact, is plainly a bystander to the original interaction. The discourse types that I have been caling ?texts,? by contrast, generaly lack both imediacy and participant transparency. When an article is writen for publication, its author has no idea who, in particular, wil be reading the magazine in which it is published; nor does she have a sense beyond an educated gues about when or in what contexts it wil be read?and this aspect of the discourse situation is something that the author understands to be the case as she is writing. 7 The same is true for any text produced for public media: comic strips, films, radio broadcasts, or novels. I wil continue to use the terms text, author, and reader to describe this kind of public discourse types, their producers, and their audiences. However, I wil frely include examples from discourse genres other than writen naratives, and wil exclude directly conversational genres, even if they are conducted in writing. Discourse As for ?discourse? itself, it too is a tricky term that means diferent things to diferent people. Scholars may use it to mean anything from ?units of language which 7 There are several interesting edge cases. Durable works originaly created for private setings may be later diseminated to a wider audience, or as was common in sixtenth-century manuscript culture (as wel as in many contemporary online communities), works may be created for semi-public distribution within a relatively wel-defined reading community. 64 are larger than single sentences? or ?language in context? (Stubbs 2001: 5) to the entire complex of symbolic activity and power relations within a given social institution, as in Foucault?s ?The Order of Discourse? (1971) and The Archeology of Knowledge (1972). In critical theory, social theory, and related branches of philosophy, the word and its variants appear most often in the context of the study of very large social structures and their origins. In this later context, ?discourse? and its relatives enjoy a wide currency in cultural and literary studies; a quick search of the JSTOR digital journal database, for example, turns up many hundreds of articles featuring the phrases ?discursively constructed? and ?discursive construction.? This usage roughly follows Foucault?s terminology and preoccupations: a ?discursive formation,? in Foucauldian terms, is a matrix of intersecting, often contradictory, ways of representing and thinking about a given set of social realities, which together actualy constitute the social realities that they represent. Critical atention is thus directed to the question of whether a particular institution, ideology, or societal construction can be held to exhibit the property of ?discursivenes?, and the implications of possesing (or not possesing) this quality. Other social-theoretical approaches use the term to describe a particular relationship betwen language and power, as in the work of Marxist theorists such as Louis Althusser (1970) and Fredric Jameson (1981) and feminists such as Judith Butler (1990), al of whom are interested in the way that prevailing ?discourses? can promote inequality and marginalize those who difer from the status quo. Here 65 ?discourse? broadly refers to an institutionalized way of thinking, speaking and writing that reinforces hierarchies by defining the limits of what is appropriate. Al of the above schools of thought refer to very large and difuse entities when they speak of a ?discourse?, and tend to focus on the relationship betwen culture-wide conventions of linguistic behavior and other features of that culture. I propose a narower and more concrete use of the term. When I say that literary texts are examples of discourse, this means I am treating them as: 1. Slices of human action and interaction, rather than bare asemblages of linguistic constructions; 2. Examples of language use, rather than simply examples of language; 3. Dynamic rather than static: built up, diseminated, and experienced over time; 4. Something that is at least potentialy collaboratively constructed; and 5. Possesed of setings. By this last item, I mean that instances of language use occur in a time and place. They invoke and are governed by an asortment of cultural norms that are specific to the time and place of their construction and reception. They have varying purposes and objectives. They have participants and culturaly-agred-upon roles that those participants can fil. Their participants understand them as instantiations of any number of distinctive pre-existing categories of activity?what Erving Goffman (1974) cals ?interaction frames?. They can be transacted in diferent media, such as speaking, gesture, sign, handwriting, print, or various mixtures thereof. Al these 66 factors are elements of the discourse seting or discourse situation of a given piece of discourse. The questions I am addresing here are: How and to what degre do the discourse activities and discourse setings of texts work acording to the same rules as conversation? How do they make use of social cognitive abilities that sem to be grounded in direct interpersonal contact in some very concrete ways, including the opportunity to se and hear another person and their surroundings? Common Objections to the Conversation Model for Texts A variety of objections are frequently raised to the idea that reading may be usefully understood as a collaborative, or conversational, activity. These fal into two broad categories: complications that arise from the separation of readers and authors in space and time, and complications that arise from the diference betwen the ostensible speaker of a literary text and the real, historical author or authors of that text. Authors and Readers are not Physically Co-Present Authors are not present during the comprehension of a text, nor are readers present during a text?s creation. This is the objection underlying Wimsat and Beardsley?s ?Intentional Falacy,? the New Textualist approach to legal interpretation, and more radical post-structuralist arguments (e.g. Barthes 1977) that the very concept of the author should be considered defunct. Concretely, there are several ways in which this lack of co-presence makes for significant disanalogies with 67 basic conversational interaction: The discourse situation of the published text means that author and reader have no shared perceptual and referential common ground in which the discourse is situated. They have no aces to the usual physical cues regarding their interlocutor?s interest, comprehension, and visual atention. Most importantly, perhaps, they have no opportunity to engage in the imediate fedback loop that is so crucial to conversation. Readers cannot ask authors for clarification; they cannot confirm their understanding or contribute new ideas. Authors cannot modify their contributions on the fly in response to an individual reader?s responses. What?s more, there is no shared interaction betwen a reader and an author leading up to the production and comprehension of a narative. Naratives in conversation normaly emerge out of the ordinary give and take of conversational turn taking; they are localy occasioned (Jeferson 1978), their progres is shaped by audience contributions over the course of many turns (Goodwin 1986), and conversation generaly continues after the narative is complete. There is rarely any comparable preliminary and consequent interaction betwen the author and the reader of a text. Who Is Speaking? Dixon and Bortolussi (2001), among others, argue that the text communication model is rendered fataly problematic by the fact that the speaker or narator of a literary text is not the author. Readers typicaly have even les aces to the intentions of an author than the mere lack of co-presence would suggest, because the real person who wrote a given text is not speaking as herself, and readers are often in a very poor position to tease apart the author from the speaker. This complication is 68 especialy obvious in the case of texts with overt narators (Booth 1961) who clearly diverge from the beliefs, felings, and biographical details of the author. However, it is also importantly true even in cases such as expository texts in which the narator appears to be identical with the author, or naratorles and dramatic texts, such as narative films, radio plays, comics, and other kinds of sequential art. Where one is available, readers often focus more on the narator than on the real or implied author of a text. Indeed, they wil often make unsupported or faulty asumptions about the author based on information provided by a narator or focalizing character. For example, readers tend to asume that the author of pasages in which perceptual information is atributed to a male focalizer is also male, and that pasages in which perceptual information is atributed to a female character were writen by a woman (Dixon and Bortolussi 2001). Graeser, Bowers, Olde, and Pomeroy (1999) suggest that even third-person narators are generaly not nearly as acesible in memory as first-person narators and other characters. Perhaps this is why readers often conflate the expresed beliefs of narators and even nonnarator characters with those of the author or implied author of a text. Take, for example, the asertions in (2) and (3): (2) Robert Frost said good fences make good neighbors. He never knew John Ames and Pery Brooks. [The Washington Post, May 22, 2005] (3) Robert Frost said, ?Good fences make good neighbors,? and he was right. [EzineArticles, January 28, 2007] As it happens, Robert Frost said no such thing; even these sophisticated consumers of texts are conflating Frost, the author of ?Mending Wal? (1915), with a character from that poem. The character himself is only paroting an already wel-established 69 adage, and the speaker, or narator, of the poem is quoting the neighbor in order to criticize him: There where it is we do not need the wal: He is al pine and I am apple orchard. My apple tres wil never get across And eat the cones under his pines, I tel him. He only says, ?Good fences make good neighbors?. It is not clear that these mistakes defeat the general claim that readers take the author of a text to be their interlocutor, and that they infer an author?s intentions as they interpret a text. Readers who are familiar with the conventions of Western-style fictional naratives do understand that the author of a text created its narator, and can draw inferences about the narator based on what they know or asume about the author or implied author. For example, knowing something about the character, publication history, and contemporary social norms of Jonathan Swift can help us to recognize the irony in ?A Modest Proposal,? and we can sek out more specific information about his expresed beliefs in trying to sort out what is and is not meant to be ironic in Gulliver?s Travels. Readers also commonly wonder whether inconsistencies in a text are deliberate or acidental, which similarly requires appeal to the real author?s communicative intentions. Where the divergence betwen the implied or apparent author of a text and the real author becomes more problematic for the text-as-communication model is in the pervasive asumption of unitary authorship. In order to read the narative as a narative, readers must act as if the communicative intentions behind the text are coherent. Yet a single work is often not?perhaps even usualy not?a unified performance of just one person?s communicative intentions. For one thing, any 70 extended narative may be the work of several years. It is hard to imagine that an author?s intentions don?t shift over the course of that writing, and the order in which various parts were writen is not at al visible to most readers. Films and television are of course the product of many diferent people?s communicative intentions: screnwriters, producers, directors, cinematographers, actors, and editors. It is frequently dificult, if not impossible, to recover the individual contributions that comprise the final product. If ordinary readers cannot and do not sort out what elements of a text were contributed by what real agent and when, surely they are not interacting in any meaningful way with those agents. This is not a fictive interaction with a real author, but a fictional interaction with an imagined author. Proposal: The Literary Event Consists of Multiple Joint Activities Pascual (2002) points out that al sorts of discourse invoke varieties of ?fictive interaction?, conceptualizations that are underwriten by a particular and pervasive conceptual blend (Fauconnier and Turner 2002) in which a situation is imaginatively reconstructed in terms of a canonical conversation frame (Filmore 1982). The result is discourse that invokes or relies on an imagined interactional structure that diverges significantly from the observable communicative situation. Certainly both authors and readers understand what they are doing at least in part by means of this fictive interaction blend. As we wil se, however, many of the concrete interactive experiences that characterize face-to-face communication do literaly, or factively (Talmy 1996) take place as part of the interpretation of texts. There are at least four kinds of intersecting joint activities involved in the 71 discourse situation of a published narative text: 1. Represented scenes of joint atention that take place betwen characters (discussed in more detail in chapter five); 2. Interactions among participants in the development and production of the published text: the credited author, editors, publishers, and so on (discussed in more detail in chapter six); 3. The relationship betwen historical authors and individual readers, sometimes mediated by their fictive counterparts; and 4. Interactions in which texts themselves serve as objects of joint atention within reading communities. I wil take the last of these first, beginning with the conversations that readers have with other readers. These interactions draw on many of the concrete resources of imediate conversation that authors and their readers cannot share, putting the text into the stream of talk and requesting or suggesting glosses, fresh interpretations, and alternate construals as they go. As we wil se, these communities of readers can then take their texts as objects of extended joint atention in order to build up new? sometimes quite radical?conventions of interpretation. Inter-Reader Interactions Collaboration in Conversation If texts genuinely and crucialy involve joint activities, we should be able to se evidence of that collaboration in their production, disemination, and reception. When speakers and hearers interact, they engage in what Clark (1996: 212) cals joint 72 construal: ?For each signal, the speaker and addreses try to create a joint construal of what the speaker is to be taken to mean by it.? Sometimes this happens smoothly and without incident, but it always requires actions on the part of both speaker and hearer. In Clark?s terminology, the speaker proposes a joint project and the hearer takes it up. The uptake response provides evidence that the hearer understood the speaker?s utterance, and also of the way that the hearer is construing that utterance. For example, if Ann says to Bil, ?Eat some spinach,? each of the responses in (4) would suggest a diferent construal of that utterance: (4) a. Yes, ma?am! [an order] b. Thanks, I?m fine. [an offer] c. What a good idea! [an advisory] If the displayed construal matches Ann?s original intentions, she can proced on that basis, which alows Bil to understand that this construal is now jointly held betwen them. If the construal doesn?t match, she can correct it (?I?m not asking you, I?m teling you!?) or leave Bil?s construal unchanged and revise her own intentions about what she should be taken as doing. Construals can be also revised or corrected over the course of multiple conversational turns, as ilustrated by the following example, taken from an exchange on an Internet discussion board (Unfogged.com 2006): (5) Revised construal 1BL: To the extent that your project here is instead just to emphasize that I, personaly, am a bad person (because your contracts profesor and I took some of the same clases in college), I disagre with you. At least in this respect. 2LB: No, no, I think you're a bad person for entirely diferent reasons. 3BL: Such as? 73 4LB: Crap, I'm such a wimp. I was considering pasting a ?just kidding! :-)!!)' on the end of that, and decided not to, but now I have to. No, I don't think you're a bad person at al, not in the slightest. I just saw a straight line and took it. Here the revised construals pertain to the afective content of the speaker?s meaning. LB?s utterance in turn 2 sems to be construed by BL in turn 3 as an insult; in turn 4, LB adopts that construal and apologizes, proposing a clarifying revision to her earlier remark. References, too, can be worked out over the course of several conversational turns, as in the following example, from the Michigan Corpus of Spoken Academic English (MICASE OFC150MU042): (6) Clarifying reference S3: yeah yeah that?s (right) and, remember like Lana Lane or some or Lana something S1: Lois Lane? [suggestion] S3: no no it?s Lois Lane but no there?s another girl Lana right? [partial rejection, partial acceptance] S7: yeah that takes over the paper [confirmation and continuation] S3: yeah [acceptance] Collaboration in Text Interpretation Authors and readers are not in a position to engage with one another in the kinds of exchanges found in (5) and (6). However, co-present readers of the same text certainly are, and this is exactly what they often do, putting the text into the stream of talk and requesting or suggesting glosses, fresh interpretations, and alternate construals as they go, using al the usual resources of face-to-face communication. Narayan (2007; in preparation) provides an excelent example of how co- present interpreters of texts make use of the resources aforded by face-to-face 74 interaction, including gestures, posture, and gaze management. In this study, pairs of subjects sat facing one another, positioned so that one participant could se panels from various comics projected on a scren, while the other could not. The task of the first participant (here, P1) was to describe each panel until both participants agred that the second participant (P2) understood what was being depicted. In this example, P1 has dificulty coming to the canonical explanation of what the panel depicts?a car pulling across trafic to nab a parking spot, while the car?s driver triumphs and the driver of the car he has cut off honks his horn in anger. P1 has repeatedly described the car as ?pulling out of a spot,? and once as ?tryna pull out into the road.? P2 is able to recognize that there is something wrong with P1?s interpretation, and over four and a half minutes and several dozen turns of conversation, the two work to revise P1?s original construal, ariving eventualy at a mutual understanding of the image that matches the canonical one. (7) Revised construal P1: The car parked behind the bug is so hard to se, could even be a taxicab, it?s just a yelow car, um.. It?s hard to se because the word honk Um Is basicaly over the top of it. P2: It?s interesting that it says SCORE, maybe what it means is he actualy got the spot, as opposed to.. looks at P1 P1: Ohhh! Yeah! Here P1 and P2 have generated much the same kind of revised construal sen in (X). P2 uses his gaze to check P2?s comprehension, prompting her to confirm that she agres and wil adopt his revised construal. While they may have no opportunity to 75 interact meaningfully with the creator of the original text, their interpretation of the text is highly interactive, and relies crucialy on cues from one another?s gestures and gaze as wel as from their talk. They are active co-participants in this interpretive conversation. Readers can also collaborate in establishing and clarifying references in an existing text. In the following example, from an undergraduate clasroom discussion on Philip Roth?s novel The Ghost Writer, a student, S4, combines quotation of the text with a variety of conversational resources to signal her own comprehension and confusion to the instructor, S1. For clarity, I have used italics to mark the words and phrases that the speakers are reading from the novel. Otherwise I have retained the transcription conventions of the original, in which punctuation indicates prosodic rather than gramatical features. (8) Clarifying references (MICASE LES300SU103) S1: um, can you start too late mother. page one-oh-six? and, read loud enough so that everybody can, hear you. S4: too late mother, didn't you read the ten questions from, Nathan Zuckerman? dear, i did posses a copy, and the leter too the big thre mama um, strikers? S1: m, they're those these are two Nazi um, officials. (actualy) [S4: and i, ] they were both, they both were involved in Nazi propaganda. S4: and your son what about um, the judge?s um, humility? where is his modesty? uh he only e- he only meant what happened, to the Jews in Europe not in Newark. we are, no- we are not the wretched of, Belsen who is Belsen? S1: Belsen uh Bergen-Belsen is a concentration camp and this is the camp where Anne Frank died. S4?s reading aloud here does far more than simply restating the words of the text. She uses intonation, pauses, restarts, and inserted requests for clarification to bring the advantages of face-to-face communication to bear on her understanding of the text. 76 S1 does not have to gues where S4 would benefit from her greater knowledge of the background information necesary to make sense of various references in this pasage?instead she customizes her contributions and clarifications based on S4?s performance. Roth?s text is being used as a discursive element reframed in S4?s own voice, alowing her to signal what she does and does not understand. Because we understand multiple performances of the same words as being representations of the ?same? text, readers can and do use both spoken quotations and multiple physical copies of a book as a way to treat the text itself as an object of joint atention. Collaboration and Change Within Reading Communities We can also step back and se these kinds of joint interpretations taking place on a larger scale, distributed over both time and space. On the basis of a shared history of many such interactions, larger groups of readers can and do form ecentric, insular interpretive communities, generating interpretations that may wel be wholly or largely impenetrable to readers outside the group. One ilustrative example of this kind of reading community is a particular set of fans of the Sherlock Holmes detective stories writen by Arthur Conan Doyle. From the first publications of Sherlock Holmes stories in the late 1880s, readers, editors, advertisers, and wags have spoken of Holmes as if he were a real person. For some of these people, this belief was purely sincere. Doyle biographer Martin Booth (1997) reports, for instance, that the British post office had received leters addresed to Holmes until at least 1950. As early as 1890, Doyle marveled that a tobaconist in Philadelphia had writen to his editor, J. M. Stoddart, to inquire ?where he could get a copy of the monograph in which Sherlock Holmes described 77 the diference in the ashes of 140 diferent kinds of tobaco? (Gren 1986: 4). Booth also tels an anecdote of his own which, even if he romanticizes slightly, indicates the widespread and continuing nature of this belief. When Sherlock Holmes came up in conversation at a hotel where Booth was lodging at Naini Tal, in the Himalayan foothils of India, ?Ah, yes!? [the proprietor] exclaimed. ?Shur-luck Homes! You know he came to Naini Tal?? ?You mean,? I corrected him, ?Arthur Conan Doyle came to Naini Tal?? ?No, sahib. It was Shur-luck Homes.? When I asked why he had come, the proprietor did not know. It had happened before the war, before he was born. His father had told him about it. (1997: xi) There is also a lively tradition of intentional ?believers? in the existence of Holmes and Watson. These fans, who cal themselves ?Sherlockians?, adopt for playful purposes the pretense that they believe that Holmes was a historical figure, and Watson his real biographer and author of the tales. They produce a vast quantity of mock-scholarly articles under this conceit. The Sherlockians comprise a discourse community in the sense proposed by Swales (1990): they have common public goals, they have mechanisms for communication and information exchange among their members, they make use of community-specific genres and specialized terminology, and members of the group share a high general level of relevant expertise. An interesting thing has happened over the course of many years of Sherlockian practice. The performance of this shared idiosyncratic construal of the texts has become routinized within the community, and as a result (Tobin 2006) it has 78 undergone some changes that look very much like the ways that the meanings of individual words can shift over time. In lexical semantics, the routinization of an expresion (Haiman, 1994; Traugott and Dasher, 2002; Hopper and Traugott, 2003) is asociated with signal simplification and semantic bleaching. For example, the English word ?goodbye? has undergone both kinds of changes since its origins in the phrase God be with you. Its form is reduced phoneticaly, and the meaning has become more abstract; some of the original semantic content of the phrase has been ?bleached? away. Similarly, as the Sherlockian community dedicated to producing and diseminating works that adopt this stance has solidified its institutional status, the explicit ironies and parodies that were the original raison d?etre of the form have faded away. As the Sherlockian style and atitude becomes entrenched in the discourse of the Sherlockian community, the ironic dimension of the stance sems often to become increasingly de-emphasized or abandoned altogether. These readings invoke a historicaly ironic frame that has been bleached of much of its ironic meaning. The task of the Sherlockian is to adopt a mixture of the methods of Sherlock Holmes and those of an academic historian to analyze the stories. Inconsistencies, omisions, and offhand remarks provide the grounds for their investigations, the goal of which is to determine the ?facts? of Holmes? life and related fictional maters. Popular topics include the question of what college Holmes atended, the name and number of Watson?s wives, the details of Holmes? drug use, and the location of Watson?s war-wound. To the uninitiated reader, the resulting esays can be dificult to distinguish from sincere confusion over the veracity of the tales. The very first 79 Sherlockian esays, however, fit clearly into a genre that was very familiar to most relatively sophisticated readers: the academic parody. A famous early example, widely cited in today?s Sherlockian community as the foundational work in the genre, is Ronald Knox?s lecture ?Studies in the Literature of Sherlock Holmes? (1911/1968). Knox?s esay is primarily a parody of the analytical style of the Biblical ?higher criticism? asociated with David Friedrich Strauss, Ferdinand Christian Baur, and Ludwig Feuerbach, among others. This approach to theological criticism sought to apply Hegelian theories of history to biblical criticism. It aspired to determine the sources of the various books of the Old and New Testaments, and eventualy to authenticate or discredit each one. The amusement value of Knox?s esay lies largely in the satirical substitution of such a frivolous subject as Sherlock Holmes for such elevated theological objects. Later Sherlockian writings put far les, if any, emphasis on spoofing the preoccupations and stylistic mannerisms of academic prose, and far more on the particulars of the Holmes texts. More and more references are internal to the extended discourse within the Sherlockian community, and markers of non-seriousnes such as hyperformality and represented intonational exaggeration are far les frequent. For example, a typical recent article in the Baker Stret Journal, Darak (2000), focuses on the question of what Holmes did with himself during the period betwen his retirement in 1903 and the publication of ?The Problem of Thor Bridge? in 1922. This subject is not a means of targeting some object of parody, as were the arguments of ?Studies in the Literature of Sherlock Holmes?. Instead, it is more importantly connected to a real, ongoing discussion taking place within the Sherlockian 80 community. The author of this article gives credit for competing theories among his felow Sherlockians, and al citations are of the stories themselves or of other analyses published in the pages of the BSJ, where Knox?s citations are patently made up and overtly parodic. This increasing seriousnes makes sense within the history of the increasing institutionalization of the Sherlockian stance and the discourse community where it might be indulged. Originaly, the community of Sherlockians was smal and intimate, and the earliest Sherlockian esays were published in setings where a sizeable portion of the audience was likely to be unfamiliar with the form. The Knox article was published at Oxford as a one-of spoof; esays by the original smal cohort of the Baker Stret Iregulars, in the 1930s, were often published in the pages of the Saturday Review of Literature, where the founder of the club, Christopher Morley, was an editor. Under these circumstances, there was limited opportunity for the stance to get heavily routinized outside of a smal circle, continualy in the position of enacting it for audiences who would find it novel. The society had only moderate institutional reality. In the 1940s, however, Edgar W. Smith took over the leadership of the Baker Stret Iregulars and began the publication of the Baker Stret Journal, which has been publishing continuously since. Today there are dozens of societies dedicated to this particular mode of Holmes appreciation. Now there is a pre-existing framework in which al Sherlockian writing makes sense; it doesn?t have to be a joke in order to avoid being cast out as an eror. For its practitioners, the stance has become unremarkable, and need not be acompanied by any markers of a sarcastic or 81 otherwise unserious atitude on the writer?s part; indeed, there is no requirement of facetiousnes for current practitioners of the form. As in ordinary conversation, the succesful production and comprehension of any one these naratives does not stand alone, but as part of a longer stream of interactions. This is al very wel as long as both readers and authors are active participants in the ongoing discourse. But as the Baker Stret Iregulars have become more institutionalized, Sherlockians have also come to inhabit multiple roles within the matrix of joint activities that I outlined earlier in this chapter. They are readers of the Holmes stories, engaged in building up shared interpretations of these stories with their felow readers. But they are also reading and producing published texts, which may be encountered by readers who do not share the Sherlockian common ground. This brings us back to the question that opened this chapter: Are readers and writers in a conversation with one another? Back to Readers and Authors Even in ordinary conversation, not al hearers are addreses. The experience of reading has several points in common with the experience of being an overhearer, rather than the addrese, of a face-to-face conversation. Authors and readers alike behave in many ways as if readers were ratified bystanders, for whose benefit the ostensible speech acts of a text are displayed. Hearers come in a number of diferent varieties: for instance, one may be an addrese, a deliberate or acidental eavesdropper whom the speaker is not aware of, or a bystander whose aces to a piece of talk is perceivable by those who are doing the talking. Bystanders can be ratified participants in an encounter (cf. Goffman 1981: 82 131-137) as when a profesor answers one student?s question during a twelve-person seminar, or not, as when diners at diferent tables in a restaurant expect one another to engage in the polite fiction that they cannot hear one another. The degre to which speakers are aware of their bystander listeners can vary, as can the degre to which they take their needs into acount. Gerig (1993: 110) proposes that readers do indeed correspond to one particular kind of non-addrese hearer commonly involved in ordinary face-to-face conversation. In his acount, both authors and readers typicaly ?behave as if readers are side-participants??not overhearers?to the discourse of a narative text, and ?in that role, authors intend readers to be genuinely informed by narative utterances.? While overhearers have to try to make sense of utterances that were produced with no atempt to ensure that they share the speaker?s perspective, Gerig argues, authors conceptualize their readers as intended, ratified participants in the narative discourse, whose needs must be taken into acount. The side-participant acount has the merit of explaining some common ways in which narative texts tend to defy verisimilitude. For example, we often se a character or narator using nominals that corespond to a much lower degre of acesibility (Ariel 1990) than one would expect given the supposed state of shared knowledge betwen the represented speaker and her ostensible addrese. Striking examples of this phenomenon are common in serial naratives, in which a mere handful of represented conversational turns may be published over the course of many days or months. The first panels of daily comic strips are full of them. (9), for example, is from a recent instalment of the soap-opera strip Mary Worth: 83 (9) Dawn: My father wouldn?t approve of my dating Drew! He wouldn?t be able to overlook the age diference! Mary: Give Wilbur more credit, Dawn! Stilted though this dialogue may be, it does some useful work anticipating the needs of many of the strip?s likely readers. Mary Worth is a daily comic with a wide variety of secondary characters. Readers, even those who frequently read the comic strip but have not been following the curent story arc, may be unfamiliar with who Dawn?s father is, despite Dawn and Mary?s knowledge. The side-participant hypothesis explains this sort of usage nicely. At the same time, there are important ways in which readers are more like overhearers of a text?s author than they are like co-participants of any sort. Overhearers have aces to the utterances that make up a stretch of discourse, but lack the opportunity to participate in the negotiation of meaning and exchange of clarifications. This mised opportunity has real efects on their ability to understand what speakers are saying, even when they can hear every word and se every gesture. Because they have no chance to collaborate in establishing a shared perspective, they are slower to comprehend a stream of discourse than direct participants in that discourse (Schober and Clark 1989). Readers of published texts are in much the same position: no mater how thoroughly an author?s comunicative intentions are directed towards generating particular interpretations for a text?s real readers, readers are at a major disadvantage in comparison to any co-present interlocutor. Similarly, authors can end up as overhearers of the interpretive conversations that go on among the readers of their own work. I am told (Lelenberg, personal 84 communication) that Arthur Conan Doyle, for example, reacted to samples of Sherlockian writing with confusion and distaste. Meanwhile, today?s popular culture supports a multitude of similarly idiosyncratic reading communities for al kinds of novels, television shows, comic books, and other texts, producing enormous quantities of collaborative interpretations and alternative ?replottings? (Gerig 1993). It is common to find the authors of the source texts expresing a certain amount of bewilderment about these discourses, even if they are ultimately flatered by the atention and sympathetic with their readers' desire to find a creative outlet. J. K. Rowling, author of the wildly popular Harry Potter series, spoke of this experience in an interview (2006). For a long time I never looked. People used to say to me, ?Do you ever look at the fan sites or se what people have said online?? I was truthful; I said I didn't. Then one bored afternoon, I googled ?Hary Potter.? Oh.. my.. God. I had NO idea. What we se happening here is a proces in which authors are neither part of nor entirely oblivious to the kinds of extended, highly participatory, interpretive discourses taking place among communities of their readers. Instead, they become overhearers of their own readers? interpretive conversations. Sumary Theoretical analyses of discourse procesing often asume that texts in general and published fictional naratives in particular should be thought of as a form of interaction betwen the author and the reader, in which writers and readers occupy the positions of speaker and addrese. While these analyses are right to look for 85 interaction in the literary event, they are wrong to sek it primarily in a direct relationship betwen author and reader. In functional terms, readers act primarily as overhearers, rather than as addreses, of authors? textual utterances. In this role, they engage in side conversations with other readers, treating the texts as the objects of their joint atention. These interactions betwen readers draw on many of the concrete resources of imediate conversation and shared visual atention that authors and their readers cannot share. Indeed, authors who encounter the product of these interactions often find themselves inhabiting the role of overhearers?even eavesdroppers? themselves. These observations do not undermine the notion that participation in literary discourse is shaped by the same kinds of interpersonal and cognitive demands as conversation. Quite the contrary: the standard objections to treating texts as examples of interactive, communicative discourse can largely be explained by taking acount of the several diferent sites of interaction that textual discourse afords. Further, as we wil se in the next chapter, the mechanics of social cognition are directly implicated in formal features of literary texts. That is, the structure of published fictions reflects the fact that authors and readers treat texts communicatively, and often exploits the peculiarities of our intersubjective minds. 86 Chapter 4: Efects of Egocentrism Social Cognition and Literary Criticism In this chapter, I take a closer look at the details of social cognition, including how, when, and how acurately speakers, hearers, and readers conceptualize the minds of others. I argue that that those mechanics are directly implicated in the reading experience. There is a growing interest in literary studies in the relationship betwen fiction and a cognitive capacity known as ?theory of mind?. These studies are just beginning to appear in print (Zunshine 2003, 2006; Herman 2006), but the subject has been a popular one of late at conferences (e.g. Mancing 2006, Gren 2006 and Bronshteyn 2006) and an entire conference dedicated to ?Theory of Mind and Literature? convened at Purdue University late last year (2007). Broadly speaking, the term ?theory of mind? is used in the cognitive sciences to refer to any ability to understand that other people have minds like one?s own, with thoughts, beliefs, desires, and intentions that may be diferent from one?s own mental states, and to the ability to hypothesize acurately about what these mental states might be. However, as mentioned in chapter two, there is a great deal of debate within cognitive and developmental psychology over what a ?theory of mind? consists of. Most researchers agre that the ability to se others as intentional agents who posses mental states is a key competence of human beings, but the mechanism of this competence is far les setled. For this reason many prefer the more neutral term 87 ?concept of mind? to describe the capacity. Some researchers have argued for a so- caled ?theory theory? (e.g. Morton 1980; Gopnik and Welman 1994; Gopnik and Meltzoff 1997). In this view, understanding the thoughts and intentions of others involves an explicit representation of other people?s mental states, based on an actual theory of how these other minds operate. This theory is built up over time, beginning with a simple model that grows more complex and refined over time. Others (e.g. Goldman 1992; Haris 1992) have argued for a simulation acount in which our ability to understand others is based in a proces of role taking: simulating others? thoughts and felings rather than theorizing about them. Many of the applications of research on concept of mind, or theory of mind, to literary studies come at it from the perspective of ?theory? theorists, and more specificaly a subset of this school of thought as sen in work by Simon Baron-Cohen (particularly Baron-Cohen 1995) claiming that the human mind has a specialized ?theory of mind module?. I do not subscribe to this view myself, but also find that this diference of opinion has only moderate bearing on most of the above-cited literary applications. These are mainly concerned with the simple fact that people do form representations of other people?s beliefs, rather than the mechanics of how these representations are constructed. Lisa Zunshine, for example, one of the most prominent literary theorists working in this area, has writen extensively on the notion that readers formulate a ?Theory of Mind? with respect to fictional characters (2003, 2006). Her work focuses on the connection betwen this Theory of Mind and the human capacity for metarepresentation: embedding clauses or representations several layers deep. In her 88 acount, the major repercussion of literature?s ability to chalenge and exploit the human readines to posit mental states can be found in the way that fiction leads to multiply embedded representations??Peter thinks that Clarisa notices that Richard wants???which grow increasingly dificult to maintain in active consciousnes. This is certainly an interesting observation connected to conceptualizing other people?s minds, but it is not one that draws greatly from the details of how that conceptualization occurs. I would like to suggest that it is very much worth looking at the details of social cognition, including how, when, and how acurately we conceptualize the minds of others, and that those mechanics are of direct interest to literary studies. I have been arguing that literature is in an important sense made out of scenes of joint atention, both in practice (the activities of the many real people involved in producing, reading, distributing, and responding to texts) and in representation (the activities of literary characters and the encoding and manipulation of viewpoint in naration). To the extent that this is so, literature is not created out of joint atention but created of it. Literature is bound by the rules and mechanisms of joint atention and, by extension, of social cognition. This includes biases in the system, the shortfals of our intersubjective talents. Even quite young typicaly developing children have striking intersubjective skils; as described in chapter 2, they generaly begin to coordinate their interactions with objects and other people in a joint atentional way by their first birthday. By the time they?re toddlers, they can expres and respond to communicative intentions in impresively sophisticated ways, producing and understanding demands, asertions, 89 and questions that reflect a flexible implicit recognition that other people?s intentions can difer from their own and may be influenced by their behavior. But young children are also substantialy worse at keeping track of what other people believe than adults and older children are. One of the best-documented limitations of young children in this regard is the dificulty they have with a group of activities known as ?false-belief tasks?. The standard false-belief task, first developed by Heinz Wimer and Josef Perner in 1983, works as follows: the subject is shown a cookie box containing pencils instead of cookies. Then they are asked what another child, shown the closed box, wil think is inside. Starting somewhere betwen 3 and 5 years of age, typicaly-developing children are able to work out that the corect answer is ?cookies.? They understand that the other person wil have a false belief about the contents of the box, a belief that they do not share. Younger children, however, fail this test and others like it, and they al fail in the same way: by claiming that the other child wil also believe there are pencils inside. A common interpretation of these findings is that young children lack a theory of mind: that they cannot predict others? behavior in this kind of scenario because they don?t yet have a concept of what a belief is, or, more generaly, because they don?t yet understand that other people have mental representations of any sort. But as we have sen, even though adults don?t have the trouble with false-belief tasks that thre-year-olds do, they too are far from perfect at keeping track of diferences betwen others? beliefs and their own. A number of cognitive, social, and developmental psychologists (e.g. Birch and Bloom 2003; Keysar, Lin, and Bar 90 2003; Birch 2005) argue that false-belief findings with children and curse-of knowledge findings with adults reflect a single fundamental bias in social cognition, one that persists across development but is more severe in children, perhaps because they have les inhibitory control than adults do (Leslie and Polizi 1998). For adults and children, alike, then, ?theories of mind? are imperfect, and the direction in which they are off-kilter is predictably self-centered. The things that are highly salient in our own minds?the tune of the song we are tapping out on the table, our own beliefs and knowledge about the topic of conversation?tend to sep into our notions of what others know or have noticed, or our notions about the obvious characteristics of objects in our environment. Because our own intentions and privileged information about the intentions of others stand out in our own minds, it can be hard to remember how much les apparent they may be to others. Certain discourse circumstances sem particularly vulnerable to substantive misunderstandings caused by the curse of knowledge. Epley, Keysar, Van Boven, and Gilovich (2004) demonstrates that time presure, for example, increases curse-of- knowledge biases. So too does the absence of paralinguistic cues such as intonation and gesture. This later fact sems to make writen discourse generaly more vulnerable than face-to-face communication. Many people have observed that electronic communication sems even more vulnerable yet (Epley and Kruger 1994; Kruger, Epley, Parker and Ng 2005). Usage guides are full of cautions against using sarcasm in email and other online conversations because writers have trouble appreciating just how ambiguous they can be in the absence of the visual, prosodic, and contextual cues of face-to-face conversation. 91 Despite these frequent warnings, people persist in overestimating the transparency of their communicative intentions, and not just in writing. Albriton, McKoon, and Ratclif (1996) found that both profesional speakers and untrained college students failed to provide sufficient prosodic cues to alow other speakers to gues which of two possible meanings they were trying to convey in uttering syntacticaly ambiguous sentences like They rose early in May. The trained speakers did manage to make the necesary prosodic distinctions when given explicit instructions about the two possible interpretations of a sentence and asked to pronounce the sentence twice, once for each meaning, but neither group did it both spontaneously and reliably. The non-profesional speakers, though, failed to produce prosodicaly disambiguated utterances for most sentences even with these explicit prompts. Given these findings, it?s not surprising that researchers who study curse-of- knowledge efects tend to treat the bias as a dangerous trap. Camerer, Loewenstein and Weber (1989: 1246) se their research as ?grounds for pesimism? about our ability to learn from either personal experience or the advice of others. Keysar and Henly (2002: 212) emphasize the ways in which our asesments of the transparency of our communicative intentions provide ?a systematic source of miscommunication.? Birch (2005: 29) concludes that ?future research needs to addres exactly how and why knowledge is a curse, and explore potential antidotes for dispeling this curse.? Indeed, the curse of knowledge does lead us to misunderstand one another, often quite badly. Literature, however, can turn this curse into a blesing. 92 The Curse of Knowledge Language use is a collaborative endeavor. In order for utterances to succed as communication, language users must engage with and keep track of the perspectives of their interlocutors. Humans are generaly excelent at divining information about other people?s knowledge, atention, and intentions, and they manage that information in al kinds of subtle ways. They try to expres their communicative intentions through actions that their listeners wil understand, and try to understand speakers by making appropriate, relevant inferences. The appropriatenes of these contributions and inferences rests not only on what is said but also on what the listener knows about the speaker?s thoughts, knowledge, beliefs, intentions, and atentions. People adjust what they say based on diferences betwen what they know and what they have evidence to believe their interlocutors know. In a clasic demonstration of this phenomenon, when Harvard student Douglas Kingsbury approached pedestrians in Cambridge, Masachusets and asked in a local acent, ?Can you tel me how to get to Central Square?? the responses he received were brief and direct, suitable for a local who could be expected to know other facts about the area, such as where the nearest subway entrance was (Krauss and Glucksberg 1977). However, when he prefaced his question with the remark, ?I?m from out of town,? the directions became more elaborate and explicit, including information about landmarks he would encounter on the way and how to recognize his destination when he got there. Kingsbury was able to elicit similarly detailed instructions when he signaled his probable lack of familiarity with local geography by adopting a rural Misouri acent. Other implicit signals of a lack of relevant expertise, such as 93 carying a map, or even looking particularly lost and puzzled, can have a similar efect. An alert traveler seking directions would do wel to keep these elements of her self-presentation in mind, even to the point of saying straight out that she?s a visitor to the city if she wants to be sure to get the out-of-town version of directions. Humans are more skiled than any other animal at discerning what others perceive, want, believe, and intend, although other mamals, and especialy other primates, do sem to understand more about intentional action and clues to others? perceptions than we once thought they did (se, e.g. Kaminski et al 2005). Humans also have unique skils of cultural collaboration: we engage in al kinds of collective activities structured by material and symbolic artifacts, as wel as complex social institutions. These cultural artifacts and cultural institutions alike rely on our ability to understand other people as intentional agents like ourselves, but with ideas of their own, who can voluntarily direct our actions and atentions to outside entities (Bloom 1996; Searle 1995; Tomaselo 1999). And yet, despite al our impresive skils of ?mind reading,? and al the ways that succesful communication requires us to perform expertly at sharing the perspectives of others, we also fal down on the job surprisingly often, and in predictable ways. For example, if we want to make ourselves understood with any degre of sophistication, we ought to avoid overestimating either how much our addreses already know or how obvious and transparent our intentions might be. That is, asuming we want to be of real help, we?d like to avoid giving the poor out- of-town visitor a terse response like ?third stop on the tramway,? when he has no idea where to find this supposed tramway, what line to take, which direction to go, 94 whether or not the driver sometimes goes past stations without stopping, how to request a stop, how to navigate the four blocks from the station to his final destination, or how to recognize it when he gets there. On a more fundamental level, we similarly want to avoid refering to things in ways that make sense to us, but no sense at al to our interlocutors; otherwise al of our conversations would be as flawed as the exchange in (10), which was produced in a study of social language in nursery school children (Krauss and Glucksberg 1977: 104). The task in this study required two people to sit on either side of an opaque barier. One, the ?speaker?, was given an ordered set of unfamiliar, hard-to-describe designs. They appeared in a transparent dispenser holding six blocks, each inscribed with one of the designs. She was also given a peg on which the blocks could be stacked. The ?listener,? on the other side of the barier, had her own peg and duplicates of the blocks, but in his case the blocks were aranged randomly on the table. The job of the speaker was to take one block at a time from the base of her dispenser and place it on the peg, then tel her partner which block she should stack on her own peg, with the object of creating two identical stacks. Adults did wel at this task, giving detailed descriptions the first time through, abbreviated increasingly as both speaker and listener became more familiar with the designs, the descriptions, and one another. Young children, however, had more trouble taking their partner?s perspective into acount in their conversational forays. Thanks to the opaque barier, neither of the four-year-olds could se the other child or the other child?s blocks, but there?s no sign of that on either side of the conversation in (10), in which each 95 contribution is meaningful to the person who speaks it, but not to anyone who doesn?t share her knowledge and perspective: (10) Speaker (refering to one of the geometric figures): ?It?s a bird.? Listener: ?Is this it?? Speaker: ?No.? Adults certainly do beter than four-year-olds at taking the perspectives of others into acount. Stil, as it turns out, they don?t outgrow these egocentric tendencies altogether. People who already know the outcome of an event overestimate what other people know about it and how easily should be able to predict it (Fischhoff 1975). They tend to believe that their internal states are far more transparent to others than they realy are (Gilovich, Savitsky, and Medvec 1998) and that others pay more atention to their appearance and actions than they realy do (Gilovich, Medvec, and Savitsky 2000, Gilovich, Kruger, and Medvec 2002). Their intuitions about the transparency of idioms like kick the bucket, to go by the board, and the goose hangs high vary acording to what they think those idioms mean? once they learn a meaning for an idiom, they tend to believe that meaning is conveyed transparently by the linguistic constituents of the phrase (Keysar and Bly 1999), while people who haven?t explicitly learned that definition don?t find it transparent at al. They underestimate how ambiguous their own utterances can be and overestimate the helpfulnes of their atempts at disambiguation (Keysar and Henley 2002). In fact, given the right circumstances, adults can be every bit as oblivious to the opacity of their communicative intentions as Krauss and Glucksberg?s four-year- olds. A nice ilustration of our foibles in this direction can be found in the ?tapping 96 study? conducted by Elizabeth Newton (1990), which has the pleasant benefit of being easy to try for yourself at home or in the clasroom. Adult participants were asked to tap the rhythm of a wel-known song and then ases how likely it was that a listener would be able to identify the song. The discrepancy betwen the expectation of comprehensibility and reality was considerable: while tappers expected that about half the listeners would be able to identify the song, the actual succes rate was only 3%. In 1989, thre economists gave a name to the bias underlying this patern of mistakes, which they described in the context of economic interest, market forces, and financial transactions: ?Beter-informed agents are unable to ignore private information even when it is in their interest to do so; more information is not always beter? (Camerer, Loewenstein, and Weber 1989: 1233). In real-life economic transactions, they observed, asymmetries in information are common. Most selers are beter informed about the true value of their products than buyers are, for example, but a knowledgeable collector of Northern Soul records may wel be beter informed about the true value of a copy of Frank Wilson?s ?Do I Love You? than the seler who found it in a box in his parents? basement. However, they found that although ?[t]he conventional asumption in such analyses of asymmetric information is that beter- informed agents can acurately anticipate the judgments of les-informed agents,? their research and that of many psychologists before them suggested that, even given financial incentives to do so and repeated fedback about the acuracy of their predictions, people were ?unable to ignore the additional information they posses?? that is, they suffered from a ?curse of knowledge.? 97 Shifting Viewpoints Two kinds of narative efects capitalize on the curse of knowledge. Zoomed out efects exploit our ability to recognize that other people are susceptible to the curse of knowledge, while remaining serenely unafected ourselves. These efects can create dramatic and structural ironies?the kinds of irony that play a central role in Tristram Shandy, Oedipus Rex, Mansfield Park, or The Adventures of Hucklebery Finn. Zoomed in efects, on the other hand, work through readers? own susceptibility to the curse. These efects serve as the source of aestheticaly pleasing narative surprises, the kind that pull the rug out from under you and leave you retracing your steps, trying to figure out how you were so misled. Both kinds of efects have to do with shifting focus betwen viewpoints in a multi-level structure of mental representations. As Michael Israel and I have argued elsewhere (Israel and Tobin 2006; in preparation), understanding irony involves a dynamic reconstrual of a slice of discourse, or a represented situation, in which atention ?zooms out? from the focused content of one mental space (Fauconnier 1985, 1997) to a higher viewpoint, from which the original viewpoint is reasesed. To achieve the rug-pull efect, meanwhile, a text must lead a reader to ?zoom in? on a lower-level perspective, losing sight of the higher-level perspective from which the eventual rug-pulling revelation wil come. First, consider how this kind of cursed perspective taking works with a very brief and relatively unsophisticated kind of story. Keysar (1994) presents a series of experiments probing the question of how people kep track of what information counts as common ground betwen characters in a story they are reading?how much 98 and what kind of atention they pay to the question of ?who knows what about whom,? in the words of one related study (Lea et al. 1998). In one of these experiments, participants read a short narative in which one character, June, recommends a restaurant to another character, Mark. After eating at the restaurant, Mark leaves a note for June saying, ?About that restaurant, it was marvelous, just marvelous.? The participants who read a version of the narative in which they?re told that Mark hates the restaurant tended to believe that June would take him to be sarcastic, while participants who read a version in which Mark loves the restaurant tended to believe that June would take him to be sincere, even though in both cases June has no aces to the privileged information about Mark?s intention. What?s more, when asked why they believed what they did about June?s plausible reaction, participants tended not to point to their knowledge about Mark, but rather to features of Mark?s utterance, such as the repetition of ?marvelous.? The vantage of the naration in this case is of the type that Genete (1980) cals ?nonfocalized,? in which the narator provides more information than any single character knows, explaining with authority both June?s and Mark?s felings about the restaurant, but with a certain flavor of ?external focalization,? to borrow Genete?s terms once again. The later kind of naration takes a purely external view of its characters, revealing nothing of their inner states. The Mark and June narative is not forthcoming in any explicit way about the characters? thoughts or intentions; instead it restricts itself to descriptions of their actions and their atitudes. 99 Mark and June are (supposedly) interacting with one another in the way that any real people would do, but the participant in the study has aces to their interactions only through the mediation of a narative that presents everything there is to be known about Mark, June, their actions, and their thoughts, as ilustrated in figure 3-1. Figure 3-1: Scope of awarenes and common ground in Mark and June narative Here we se that the scope of Mark?s awarenes encompases his own actions, thoughts, and intentions, as wel as many but not al of June?s narated experiences. June?s awarenes, similarly, encompases her actions, thoughts, and intentions, and a subset of Mark?s actions and experience. The scope of the narative, meanwhile, 100 encompases everything in the pasage, which includes everything that we can know about what either one of the characters knows, plus some information that one or the other of the characters is not aware of. And because Mark and June only exist insofar as they have been narated, the whole of what they ?know? is entirely contained within the larger purview of the narative as a whole. Competent readers are supposed to be able to keep track of both the boundaries of these perspectives and their relationship to one another; Keysar?s study demonstrates that they don?t always do so. The relationship betwen the perspectives that a reader has to navigate in interpreting this brief narative?Mark?s, June?s, the reader?s, the narator?s?can be represented in terms of the ?mental spaces? (Fauconnier 1985, 1997; Cutrer 1994) that a reader might construct in constructing an interpretation of the text. Mental spaces are structured mental representations that can be reflected in and prompted by linguistic structure, ?partial structures that proliferate when we think and talk? (Fauconnier 1997: 11). Mental spaces have contents, consisting of one or more entities, or participants, each of which can be asociated with one or more roles (for example, ?mother?). Frames, schematic understandings of aspects of the world and its workings, provide structure for these contents. For example, when a financial transaction is under discussion, speakers can refer to ?the seler? with confidence that others wil understand the reference with respect to a generaly understood model of the typical participants, roles, and dynamics of such a transaction. A space also has status with relation to other spaces. These relative statuses can be hierarchical: mental 101 spaces can contain other mental spaces. They can also be epistemic: one space can have the status of fact with respect to another space, for example, or prediction. They can be temporal: a space can be figured as past with respect to one space and present with respect to another. Finaly, mental spaces are also potential objects of joint atention, as speakers and hearers try to coordinate their mental representations and share atention to various aspects of those representations. These spaces are linked together in networks that are built up dynamicaly in working memory and can be stored in episodic memory. One mental space leads to another, and mental spaces can inherit structure from other spaces. These networks can be configured in a variety of diferent ways: event-chaining configurations that represent the conceptual structures invoked by the expresion of tense, aspect, or causation; world-structuring configurations involved in the conceptualization of beliefs, possibilities, and stories; narative configurations involved in understanding narative embedding or fre indirect discourse; frame-structuring configurations such as those involved in analogical thinking, and so on. The chain of (some of the) mental spaces and inherited structure involved in a narative like the one presented in the Mark and June task would look something like figure 3-2. Solid lines represent the trajectory of conceptual structures that are inherited ?correctly?: it is not a mistake by any lights to asume that an author knows what her characters know. Dashed lines represent the ?ilusory? propagation of information. 102 Figure 3-2: A narative configuration of mental spaces Each of these spaces contains a representation of the knowledge and atitudes of some player in the narative situation. These mental spaces, like any others, wil inherit some of their content from other spaces in the network. Propositions such as The food at the restaurant Venezia is bad can thus propagate through the network. What we might cal ?non-ilusory? transparency, then?the unproblematic asumption that, for example, an author must know whatever she tels us that one of her characters knows (otherwise, how would we ever have heard about it?)?is the projection of structure up the chain. ?Ilusory? transparency?projection that sems to arise from the curse of knowledge?is the projection of structure down the chain. 103 Though Mental Spaces Theory does not claim to model cognitive biases like the curse of knowledge, it does have a theoretical apparatus to acount for other ways that structure tends to flow from one mental space to another. Structure is projected automaticaly down the line via a principle caled space optimization: ?relevant structure not explicitly contradicted is inherited within the child-space? (Fauconnier 1997:112). For example, the statement ?Sarah?s so nice, I wish she were my sister? sets up two spaces: the Base space of the speaker?s reality, and a ?wish? space in which Sarah is the speaker?s sister. Relevant properties of Sarah that obtain in the Base and which are not dependent on the familial connection at isue?for example, Sarah?s appearance, age, and nicenes?transfer to the wish space even though they are not themselves explicitly wished-for qualities. Structure is often projected automaticaly in the opposite direction as wel, through a principle caled presupposition float. Some words and gramatical structures cary presuppositions; for example, both (11) and (12) presuppose that there is a king of France, thanks to the use of the definite description ?the King of France?. (11) The King of France is bald. (12) The King of France is not bald. Other presupposition triggers include factive verbs such as realize, change of state verbs such as stop, clefts and pseudoclefts, non-restrictive relative clauses, and iteratives. Presuppositions hold up under negation, as ilustrated by the King of France example, but it is more dificult to predict when presuppositions of embedded 104 clauses wil give rise to presuppositions of an entire sentence, or of a multi-sentence piece of discourse (se Levinson 1983:191-225). For example, the definite reference in ?Mary wanted the King of France to visit her? does normaly give rise to an interpretation including the presupposition that there is a king of France, but this presupposition can be canceled in a way that doesn?t work when the reference occurs in a non-embedded clause. Hence (13) is felicitous, while (14) is not. (13) Mary wanted the King of France to visit her, but there is no King of France. (14) #The King of France is bald, but there is no King of France. In Mental Spaces Theory, the diference betwen these cases is explained by the following rule: A presupposition floats up until it mets itself or its opposite (Fauconnier 1997: 61). In other words, the default state of afairs is for certain kinds of structure to propagate through a network of mental spaces, only stopping when it runs into a space where the contents actualy contradict it. In addition to projecting contents from one space to another, we can shift our attention within these configurations, moving our viewpoint from space to space. Every mental space configuration includes a Base, a Viewpoint, an Event and a Focus, although a single space can serve as more than one of these at the same time. The Base space serves as the subjectively construed ground of interpretation. The Viewpoint space is the space from which conceptual content is acesed. The space in Focus is the space on which atention is concentrated. The Event space is the one in which an event takes place. 105 Our awarenes of and susceptibility to the curse of knowledge can be manipulated to prompt particular kinds of shifts of viewpoint with respect to these other spaces. These shifts create distinctive experiences of surprise, suspense, and irony. Zooming In: Imersive Fiction and the Narrative Rug-Pul The Twist Ending The Mark and June example presents a clasic curse of knowledge situation, in which the ?cursed? reader has privileged information and forgets that other people might not have aces to that information. This section addreses a slightly more complicated variation on this scenario, in which the afected reader or viewer is presented with what wil eventualy turn out to be incorrect information which she forgets is asociated only with a particular, limited viewpoint. The revelation that this was an eror, that the facts that the audience thought obtained at the base level of a narative were only the belief or representation of characters within that narative, can create a very satisfying sort of surprise. At its best, this kind of twist pulls the rug out from under the hoodwinked reader; some of her most basic asumptions about the truth of what?s going on are transformed in a sudden revelation, requiring the re- evaluation of many of her interpretations of previously represented events. In the famously rug-pulling movie The Sixth Sense, for example, the young protagonist, Cole, has the unique ability to se dead people, who walk unsen among the living. For most of the movie, he struggles with this second sight with the help of 106 a sympathetic child psychologist, Malcolm Crowe, who is struggling with demons from his own past. The revelation that Dr. Crowe is one of the dead people that only Cole can se is the major surprise of the film, requiring na?ve, first-time viewers to re- ases many of the events of the film in ways that difer radicaly from their original apparent significance. To drive home both the surprise and the fair play, this revelation is followed by a rapid replay of the many moments when Crowe?s ghostly qualities were on display, but (the movie presumes) overlooked, so that the intended viewer can appreciate both aspects of the rug-pul at once. To achieve this efect via the zoomed in gambit, a narative must seduce readers into staying imersed in the story to a fault, locked into a relatively limited perspective and projecting the asumptions of that perspective through the entire mental space network. If the gambit works, the reader fals prey to the curse of knowledge, taking what she knows, or thinks she knows, and asuming that it is just so, that this view of events is coextensive with the whole story, and failing, at least for the time being, to alow for the possibility that the perspective she?s been presented with isn?t the only possible one. The first step, then, is to establish a particular viewpoint unambiguously as belonging to some individual within the world of the text, and then to entice the reader to adopt this perspective as her own. The structures of language itself encode perspective in a variety of ways. For example, some lexical items, such as tomorrow, later, or upstairs, incorporate a particular vantage in space or time as an inherent part of their meaning, but there are many other ways, as wel, in which linguistic structures reflect a particular construal of a given scene (Langacker 1987). You might 107 say that lexical and gramatical structures reveal a particular viewpoint, or that they betray it: it is impossible to expres much of anything without imbuing that expresion with a particular perspective. Consider, for example, the diference betwen (15) and (16): (15) The lanes diverge at Sixth Stret. (16) The lanes merge at Sixth Stret. In Langacker?s terms, these two expresions represent a diference of perspective involving mental scanning. The two sentences describe the same scene; their diference resides in the direction of the conceptualizer?s mental path through that scene. The perspective invoked by a given utterance can also vary acording to whether the conceptualizing agent?or focalizer, to use the corresponding term from naratology?is or is not represented explicitly (Langacker?s objective vs. subjective construal). In both (17) and (18), for example, the position of the squirel ?high above? invokes a vantage point close to the ground below; in (17) the vantage holder is construed objectively?from the outside?while in (18) it is construed subjectively, in that we are simply invited to consider the squirel from this vantage ourselves. (17) The cat spotted a squirel perched high above. (18) A squirel perched high above. (18) thus sems more readily to invite its reader to identify directly with the perspective with respect to which the squirel is ?high above?. But the reader of either 108 one may employ what Brian MacWhinney (2005) cals the ?enactive mode? of procesing, in which she adopts the presented perspective as her own. What?s more, it sems that people are more likely to proces utterances ?enactively? when they occur in the context of an extended narative discourse. ?The longer and more vivid our experiences,? MacWhinney observes, ?the more they stimulate enactive proceses in comprehension.? To reiterate: 1. Narative offers a range of possibilities for conveying embedded perspectives, such as the viewpoint of a self-conscious narator, or the reported/quoted perspectives of non-narator characters. 2. The more an embedded perspective dominates the narative, ? the more the addrese wil align her perspective with the embedded perspective; ? the more the addrese wil be afected by the Curse of Knowledge, that is, by the additional information she posseses but should ignore when imagining what others think; and ? the more the addrese wil be vulnerable to a narative rug-pull. The narative trick of pulling the rug out from under the addrese requires a masterful balance if it is to be satisfying: the rug-pulling revelation must be genuinely surprising, but it?s no good if the narative simply tels you one thing, then another. A narative that sems merely inconsistent may indeed make the reader fel like the rug has been pulled out from under her, but on a diferent level?not intentionaly, not 109 pleasurably, but through a failure of the text. The rug-pull that satisfies as a rug-pull builds up a set of expectations only to undermine them in a flash, while maintaining a sense that the undermining has al been done in a spirit of fair play. Specificaly, a rug-pull is succesful if 1. it is unexpected; and 2. it does not, in retrospect, conflict with the information otherwise presented. The Murder of Roger Ackroyd Ordinarily, we are not supposed to bamboozle our friends and expect them to enjoy and applaud our eforts, but clasic ?clue-puzzle? detective fiction aims at precisely this goal. One of the features that distinguishes the clue-puzzle mystery from the broader category of crime fiction, as wel as from sister and predecesor genres such as the thriler or the sensation novel, is a particular pair of constraints on what counts as appropriate, ?fair,? or cooperative. These two constraints and the tug of war betwen them have long been recognized, described decades ago by eminent practitioners of the form: The body of the work should be occupied with the teling of the story, in the course of which the date, or ?clues,? should be produced as inconspicuously as possible, but clearly and without ambiguity in regard to their esentials. The author should be scrupulously fair in his conduct of the game.? (Freman 1924) The reader must be given every clue?but he must not be told, surely, al the detective?s deductions, lest he should se the solution too far ahead. Worse stil, supposing, even without the detective?s help, he interprets al the clues acurately on his own acount, what becomes of the surprise? How can we at the same time show the reader everything and yet legitimately obfuscate him as to its meaning? (Sayers 1929) 110 The big surprise in The Murder of Roger Ackroyd is that the narator, Dr. Sheppard, is the murderer. At the end of the novel, Sheppard himself crows over the narative?s compliance with both of the genre?s prime constraints on fair play: I am rather pleased with myself as a writer. What could be neater, for instance, than the folowing: ?The leters were brought in at twenty minutes to nine. It was just on ten minutes to nine when I left him, the leter stil unread. I hesitated with my hand on the door handle, looking back and wondering if there was anything I had left undone.? Al true, you se. But suppose I had put a row of stars after the first sentence! Would somebody then have wondered what exactly happened in that blank ten minutes? In this case, the surprise goes against not just one, but two sets of expectations. One arises from structural consequences of the fact that Sheppard is the narator. His embedded perspective dominates the narative, encouraging the reader at every turn to align her own perspective with it. What?s more, there are conventions of the genre that reinforce this reading. Sheppard is also serving as the detective Poirot?s sidekick and amenuesis. That is to say, he is not just any narator, but a temporary Watson, making him one step further imune from suspicion. Indeed, these factors led to a certain amount of disagrement over whether this maneuver realy did count as fair play. ?The trick played on the reader in The Murder of Roger Ackroyd is hardly a legitimate device of the detective-story writer? (Wright 1927) 111 ??[T]his opinion merely represents a natural resentment at having been ingeniously bamboozled. Al the necesary data are given.? (Sayers 1929) This dispute depends on the degre to which the reader is able, after the fact, to take back the conflation of narator with author. There is something especialy unsetling about the revelation that a trusted narator has been unreliable. It is so easy to succumb to the curse of knowledge in this way that it happens invisibly; readers felt not just fooled, but downright betrayed by the way Christie had disguised her most crucial clues. This kind of twist ending can thus be especialy exhilarating or especialy infuriating (or simply tiresome, depending on how often one encounters the same technique). Perhaps this quality is the reason that the ending of Roger Ackroyd could fel so fresh and daring, even though similarly structured revelations were common enough in other genres, Poe?s ?The Tel-Tale Heart? (1843) being perhaps the most famous example. Another reason that Christie?s version of this twist was so efective and surprising, of course, was the way that it played on other conventions of the genre. One way that this kind of surprise can be deployed is by exploiting the range of narative conventions for lively exposition. Authors and readers have a contractual arangement acording to which authors must provide exposition, but without being tedious. The conventions that have arisen to acommodate this requirement provide a convenient opportunity to set readers up to believe a piece of information is ?true? at the base level of a narative, when in fact al that they know is that one character believes it to be true, or has said that it is true. This technique provides abundant opportunities for narative twists that capitalize on the curse of knowledge, since it is 112 when we overpopulate possible viewpoints with what we know (or think we know) that we are vulnerable to these sorts of rug-pulls. Many kinds of fiction are popular for their presentation of rich detail about an unfamiliar world and its workings. This world may itself be fictional, but need not be; James Clavel?s Shogun or Patrick O'Brien?s Aubrey/Maturin series are widely aclaimed for their atention to the mechanics and trappings of their historical setings. But regardles of whether the information in question is historical, technical, fantastical, or science-fictional, the Moby Dick technique of simply inserting pages of supplementary material is often viewed as intrusive (deliberately so, in some experimental works) or old-fashioned. Esayist and science fiction writer Bruce Sterling (1996) refers to these insertions of lengthy ?info-dumps? in the authorial voice as ?stapledons,? after British philosopher and novelist Olaf Stapledon, whose seminal works of speculative fiction are more widely admired for the scope and innovation of their ideas than the dynamism of their prose. But if a novel?s plot hinges around an intricate technicality of Victorian inheritance law, the behavior of Chinese ghosts, or the tensile strength of nanobot- engineered diamond, this background must be conveyed to the reader somehow. One common technique is to have characters discuss the facts with one another, what Sterling cals ?As You Know Bob? or ?maid and butler dialogue?. This approach can be managed more or les deftly, of course, and les skilful examples are often the subject of comment and parody. Regardles, this convention is deployed so frequently that readers grow very much used to the idea that lots of information about the basic facts of the seting wil arive in the mouths of characters. 113 Similarly, the narator of Roger Ackroyd fils a familiar role, that of the faithful Watson character. This character serves as the detective?s asistant and amanuensis, and as the reader?s proxy. Watsons abound in clasic detective fiction, from the prototype unnamed narator in Poe?s Dupin stories, to Watson himself, and Hercule Poirot?s own usual sidekick Arthur Hastings. The Watson character serves his own conventional purpose: He gives the reader aces to al the relevant clues, and to tantalizing hints about the detective?s superior lines of deduction, without revealing too much about their conclusions until the proper time. This convention reinforces the zoomed-in efect; while one can often overcome the curse of knowledge given the opportunity to consciously reflect on the possibility that it might be a mistake, the conventional role of the Watson character helps to discourage readers from doing so. Zooming Out: Irony and the Curse of Knowledge The flip side of the zoomed-in phenomenon is the range of literary efects that arise from the representation of the curse of knowledge and its consequences. Our familiarity with the tendency of others to fal into the traps of these egocentric biases can be exploited to create a sense of ironic distance. The presentation of fataly clashing viewpoints, constructed around curse-of-knowledge failures on the part of characters, prompts the reader to ?zoom out? to a higher-level perspective, from which the lower-level viewpoints may be sen ironicaly. 114 The kinds of phenomena that have been described as varieties of irony are so varied and numerous that it is not always clear whether they have anything in common at al. In cognitive science, it is common (Sperber & Wilson 1981, 1998; Clark & Gerig 1984; Kreuz & Glucksberg 1989), to restrict studies of irony to ?verbal ironies? such as blame-by-praise (cf. Knox 1961), or cases such as ?This is just lovely,? said biterly of something particularly squalid. Sarcasm is the paradigm case of concise verbal irony, and Swift?s ?Modest Proposal? is the clasic extended example. Verbal ironies typicaly can be ?decoded? by understanding that the speaker?s position and the speaker?s sarcasticaly adopted position difer in crucial ways. Swift?s narator proposes that Irish babies should be bred and slaughtered as meat for human consumption. The succesful interpreter understands that the implied Swift himself proposes no such thing; instead he is presenting a savage and satirical criticism of the cruelty of the English landlord clas. In other domains, irony may more frequently refer to a peculiarly sophisticated or detached atitude, such as the ironic enjoyment of camp (cf. Sontag 1964) or the simultaneous appreciation of several mutualy exclusive explanations of the world that Schlegel described as the ironic mental posture of Romanticism. Situational ironies, meanwhile, hinge on the direct opposition of actual outcomes to intended outcomes or the means by which they are pursued. An action intended to prevent some circumstance actualy brings it about, or an action intended to bring about some circumstance directly prevents it, as depicted in O. Henry?s ?Gift of the Magi,? in which a young wife sels her hair to buy a chain for her husband?s watch, while he has sold the watch to buy her combs for her long hair, or Somerset 115 Maugham?s ?Appointment in Samara,? in which a young man fles the figure of Death only to end up in the city where their appointment had been scheduled al along. Dramatic ironies arise from critical disparities betwen a character?s limited knowledge of his situation and the reader?s or audience member?s greater understanding, giving Huck Finn?s ?Al right, then, I'll go to hel? a special significance and poignancy, and adding a special frison to watching Oedipus woo Jocasta. Al of these kinds of irony have in common a particular kind of interpretive experience or atitude that comes from a doubled viewpoint, a sense that one has ?stepped back? or zoomed out from one viewpoint to another, more sophisticated view, from which one can gaze, smugly or sympatheticaly, down upon the original. Wayne Booth describes the experience in this way: ?the proces is in some respects more like a leap or climb to a higher level than like scratching a surface or plunging deeper. The movement is always toward an obscured point that is intended as wiser, witier, more compasionate, subtler, truer, more moral, or at least les obviously vulnerable to further irony. (Booth 1974:36) The experience of irony in the sense described here arises when an expresed proposition conflicts with the content of a focused space in a way that leads the conceptualizer to adjust the entire mental space configuration. In order to count as ironic, an expresed proposition in some focus space must conflict with the content of an implicit or presupposed proposition in a higher viewpoint, as ilustrated in Figure 3-3. 116 Figure 3-3: Zooming out This means that irony depends on the availability or construction of a new Viewpoint space from which one can re-aces a prior Focus space and its Viewpoint at the same time, creating a ?view of a viewpoint?, or distinguishing betwen what John Haiman (1998:80) cals ?the diference betwen a behaving and a scrutinizing self.? The proper kind of clash at a lower level of a narative configuration can prompt a reader to ?zoom out? to a higher level of the narative discourse structure. This zooming out both provides the experience of ironic ?distance? and, often by tapping into features of the existing discourse situation, a sense of complicity betwen the interpreter and some real or imaginary interlocutor. Figure 3-4 ilustrates a simple zooming-out scenario for a clasic example of verbal irony. ?Nice weather we?re having,? says one person to another, as dismal rain pours down upon them. In order to 117 achieve the ironic interpretation of this statement, the hearer constructs two spaces: a Focus space with the proposition that the weather is nice, and a new Ironic Viewpoint, which is distinguished from an ordinary observation that the weather is, in fact, not nice, by being set up as a view of the pretended or represented view that the weather is nice. This viewpoint space represents a new ground and a new potential common ground for communication betwen the interpreter and some real or imaginary interlocutor: the object of their joint atention. Figure 3-4: ?Lovely weather? The zoom-out efect of irony in this way involves a kind of alienation from the participants asociated with the lower-level space. The higher-level common ground from which one views those participants, meanwhile, is often already built into the discourse situation or genre in which the irony is presented. Since the 118 discourse situation of a novel operates on at least thre levels?the level that a reader shares with an implied author, the level of the narator or focalizer, and the level of the naration itself?the ironic zoom-out puts the reader freshly in line with the implied authorial viewpoint, providing a potential sense of complicity with the author. Because of these pre-established higher-level viewpoints and the way that information spreads through a mental space network, depictions of narators and characters faling prey to the curse of knowledge provide an excelent means of prompting readers to zoom out in this way. The resulting ironies may be subtle or broad, gentle or cruel, nuanced or coarse, depending on the manipulation of the character?s cursed thinking, its context, and whether or not the character ever awakens to the biases that serve as the basis for generating the ironic efect. To ilustrate how this works, let?s turn first to a curse-of-knowledge irony in which the character faling prey to the curse is sympathetic, but stil very much the victim of the resulting irony and the butt of a joke that the implied author and perceptive readers share. This example comes from a children?s novel by E. Nesbit, The Story of the Treasure Seekers (1899). The novel is narated by one of six siblings, Dora, Oswald, Dicky, Alice, No?l, or Horace Octavius. Very early on, this narator observes, ?It is one of us that tels this story?but I shal not tel you which: only at the very end perhaps I wil. While the story is going on you may be trying to gues, only I bet you don?t.? For the atentive reader, however, this mystery doesn?t last long. Excerpts (19), (20), and (21) come in quick succesion over the first few pages of the book: 119 (19) It is one of us that tels this story?but I shal not tel you which: only at the very end perhaps I wil. While the story is going on you may be trying to gues, only I bet you don?t. (20) Dora and Dicky did not look pleased, but I kicked No?l under the table to make him hurry up, and then he said he didn?t think he wanted to play any more. I told No?l to be a man and not a sniveling pig. (21) ?Whichever it is,? [No?l] added, ?none of you shal want for anything, though Oswald did kick me, and say I was a sniveling pig.? Once a reader appreciates that the narator is clearly Oswald, despite his atempts to hide it, and his oft-demonstrated conviction that he is hiding it, the conceit provides a good deal of humor and dramatic irony for the remainder of the novel. Every time the narator makes flatering remarks about Oswald?s bravery, kindnes, and, above al, modesty, as he does frequently, the reader can enjoy another chuckle at his expense. Perhaps one of the most important conventions at work here is the convention that the characters in a novel are, from their own point of view, ?real??as a participant in the events of the novel, the first-person speaker of these utterances has the same status as the other characters he describes. As a self-conscious narator, he is subject to al sorts of expectations basic to the conventions of a particular discourse type, many of which are general to our basic model of ordinary conversation. For example, we expect that, in the absence of any explicit signal otherwise, the same speaker wil obtain across the entire narative. We expect, too, that we can use general Gricean standards of cooperative language use to guide our interpretation of his utterances: we take him to be 120 consciously taking his audience into acount, to have the intention for that audience to recognize his meaning and his intention that they recognize that meaning; we expect him to adhere (or at least believe himself to be adhering, or intend for his audience to believe him to be adhering) to principles of relevance and quality; and so on. But the conventions of the genre also tel us that we have appeal to a higher common ground, and that there is a level of naration from the point of view of which the narator, characters, and events of the story are mere fictions, subject to the intentions of the implied author. Convention provides ample basis for readers to situate the narator Oswald in a narative situation that includes an implied audience and the asumption that the focalizer is articulating the narative for the purpose of making his intended meaning evident to that audience. So when Oswald declares that he plans to conceal his identity, and proceds to make several utterances that clearly indicate which of the six children he must be, we are aware not only that Oswald has made a ?mistake? but also that there is a more fundamental level on which these moves are not a mistake at al. A reader who is enjoying the ironic interpretive experience here is aware of the ways in which what she understands about the discourse situation diverges from what Oswald intends her to get. His vision of his addrese is diferent from that of the ?real? or Nesbit-implied reader. If we take it that the implied Nesbit intends for her reader not only to construct model 2, as ilustrated in Figure 3-5, but to recognize the clash with model 1, partaking of this view of a viewpoint provides an experience of ironic distance, and aesthetic, perhaps even moral, pleasure. 121 Figure 3-5: Clashing models in Treasure Seekers This kind of ironic construction centered on a character?s inability to distinguish betwen his or her own beliefs and those of others is not rare, nor need it be deployed as broadly as it is in Nesbit. Consider, for example, the ironies of egocentrism at the heart of Jane Austen?s Ema. Ema is a novel in which self- awarenes slowly dawns upon a character whose delightful, but fataly limited, view of events is for the most part the focalizing perspective of the narative. How is it that Austen imerses, or sems to imerse, readers in a perspective whose most striking quality is its lack of self-awarenes, while alowing the same readers to appreciate both that lack and its eventual abeyance? The answer, again, is to play on the curse of knowledge in order to prompt the reader to ?zoom out? to a higher-level perspective. Readers can then shift betwen this ironic viewpoint, where they stand with the implied author to cast a critical eye on Ema?s view of her own actions, and the sympathetic and subjectively construed perspective of Ema herself. 122 A fine ilustrative example of the way that Austen makes use of the curse of knowledge to create ironic distance can be found in the episode of Mr. Elton?s love note. At this point in the novel, the ?handsome, clever, and rich? Ema Woodhouse has befriended Hariet Smith and taken her under her wing. Hariet is swet and prety, but possesed of limited talents and inteligence. Hariet?s parentage is also unknown, and her social standing is therefore murky at best. These facts do not stop Ema, who is stoutly loyal as wel as frequently unwise, from embarking on a series of il-fated schemes to prove Hariet?s worth and mary her off to a man suitable to what Ema believes to be her proper station in life, a belief that, as it turns out, no one but Ema wil share. It is to this purpose that Ema begins an extended campaign to bring Hariet into frequent contact with the gentleman Mr. Elton, with Ema herself on hand as facilitator and chaperone. To Ema, the import and object of these interactions could not be plainer. To Elton, meanwhile, they are equaly obvious?but not the same. He knows his enthusiasm and compliments are directed at Ema, not Hariet, and as Hariet is not even a socialy viable object for his afections, he ses no ambiguity. At last, in typical Regency fashion, he sends a coy, lightly coded ?charade? of a love note, which does not explicitly name the object of his afections but should make her identity abundantly clear to anyone with any acquaintance with the afair. Ema reads it: She cast her eye over it, pondered, caught the meaning, read it through again to be quite certain, and quite mistres of the lines, and then pasing it to Hariet, sat happily smiling, and saying to herself, while Hariet was puzzling over the paper in al the confusion of hope and dullnes, ?Very wel, Mr. 123 Elton, very wel indeed?. This is saying very plainly, ?Pray, Mis Smith, give me leave to pay my addreses to you.?? ?Thy ready wit the word wil soon supply. Humph-Hariet?s ready wit! Al the beter. A man must be very much in love, indeed, to describe her so. Figure 3-6: ?Thy ready wit? Despite al other evidence to the contrary, Ema resolves the reference of thy ready wit in light of her own understanding of the relationship at hand, reliant on her conviction that everyone shares and is aware of her intentions regarding Hariet and Mr. Elton. Because she asumes that everyone else shares her belief that Elton and Hariet are wel suited to one another, and that they make a likely match, the note must therefore be addresed to Hariet. If the details of its description difer from 124 other facts, such as Hariet?s relative verbal dexterity, those discrepancies must be explained away in some other way. Elton, meanwhile, has composed his leter with the full expectation that his own intentions and understanding are just as widely held as Ema believes hers to be. Both writing and reading are thus ?cursed?, as ilustrated in figure 3-6. Both ways of interpreting thy ready wit are vexed from the point of view of succesfully coordinating Ema?s and Elton?s comunicative intentions. Fortunately, the reader is not bound to choose one to the exclusion of the other. Instead, by re-construing the events from a higher-level viewpoint, as in figure 3-6, she can interpret thy ready wit as a source of confusion, and evidence of Ema?s lack of self-awarenes. The presentation of fataly clashing viewpoints, constructed around curse-of-knowledge failures, thus prompts the reader to zoom out to a loftier perspective, and ironic distance ensues. By artfully managing this experience in conjunction with the ?zoomed in? perspective of Ema?s own subjectively construed experience, and a progresion of increasing confluence betwen the two, the novel creates a nuanced portrait of a heroine both sympathetic and flawed, and a convincing arc of growing self-knowledge and character. Sumary The dynamics of joint atention and perspective taking are crucial to the pragmatics of both conversational and literary discourse. However, they do not make communication a proces of perfect coordination. Indeed, it is riddled with egocentric 125 biases such as those arising from the curse of knowledge. As we have sen, complex narative discourses also incorporate these interpretive biases, and the apparent trap of these biases provides rich material for creative narative and literary aesthetic efects. The curse of knowledge gives rise to two kinds of literary efects: 1. Zoomed-In Efects. Texts can encourage readers to over-generalize from a limited perspective. Narative texts, especialy, offer many opportunities to present embedded perspectives, such as the reported beliefs of non-narator characters, or the viewpoint of a self-conscious narator. The more extended the reader?s exposure to an embedded perspective, the more likely she is to align her own viewpoint with that embedded view, and then to fal prey to the curse of knowledge, failing to discount this additional information when imagining what others think. Texts can take advantage of this tendency to engineer ?rug-pull? endings, surprising readers with information that contradicts the over-generalized propositions. These surprises qualify as satisfying twists because they are both unexpected and, in retrospect, consistent with the information otherwise presented in the text. Conventions that originaly evolved to fulfil other goals, such as verisimilitude, alowing the reader to be privy to a detective?s proceses but not his conclusions, or circumventing tedious exposition?can be exploited to further facilitate these efects. 2. Zoomed-Out Efects. Texts can prompt readers to recognize characters? susceptibility to the curse of knowledge, showing them to be poor judges of 126 the transparency of their own communicative intentions. The discrepancy betwen the way the characters understand their own situation and what the reader is given to understand the true situation to be is a species of dramatic irony. In recognizing the source of this discrepancy, readers ?zoom out? to a more distant perspective on the characters? perspectives. A perspective that was first construed subjectively is in this way newly construed as an object of conceptualization. The ironic zoom-out puts the reader in line with the implied authorial viewpoint, providing a sense of complicity and the feling of ?looking down on? the object, or victim, of the irony. 127 Chapter 5: Puzzles of Joint Atention in Poe and Wolf New Readings Discussion of joint atention in the preceding chapters has focused largely on the ways that literature engages directly with the social-cognitive life and predilections of its readers. In this chapter and the next, I wil turn to the question of how these concepts from cognitive science and social psychology can contribute to one of the primary goals of literary criticism: opening up spaces for new and interesting readings of individual texts. The work of interpretation is central to literary studies. Can al this talk of joint atention and social cognition add anything novel to the interpretive project, or is it just a mater of providing new terminology for old ideas? Why anyone should care about the ?social cognition? of fictional characters at al? Being made of text rather than bone, blood, and neurons, they needn?t suffer from the same cognitive limitations as those who would read or write the naratives that describe them. Indeed, they do not cognize at al, socialy or otherwise. In this chapter I propose that the concept of joint atention opens the door to fresh interpretations of two texts: Edgar Alan Poe?s ?Murder in the Rue Morgue? and Virginia Woolf?s To the Lighthouse. My major claim is that that a frequently cited afinity betwen two historical genres?the so-caled ?stream-of-consciousnes? novel of British Modernism and detective fiction of the same era?can be traced in 128 part to similarities in the ways that two famous examples of these genres deploy joint atentional scenes. Armed with a theory of joint atention, we find in these texts not just a shared interest in the relationship betwen visual atention and mental states but shared methods for their depiction. Poe and Woolf might sem to have litle in common beyond lasting critical atention and a tendency towards stylistic pyrotechnics. But looking at joint atention in these texts reveals facets of each author that might otherwise be obscured. Poe turns out to be a theorist of consciousnes, not just a psychological writer in general, but a social psychologist in particular. His atention to tracing the routes of others' thoughts displays a remarkably modern preoccupation with intersubjectivity and its material mechanics. Stripped of the fustian garb of his prose, ?The Murders at the Rue Morgue? emerges as a surprisingly modernist work. Meanwhile, Woolf?s interest in the desperate and mostly doomed atempt to understand and connect to the minds of others is so surpasingly evident as to be almost not worth mentioning at al. Similarly, it is patently obvious that To the Lighthouse evinces a pervasive interest in seing, modes of seing, and the relationship betwen knowledge and perception. Indeed, its central character is a painter, and one of its most memorable pasages centers on her atempts to ?think of? and ?se? a ?kitchen table when it?s not there.? But the similarities betwen her novel and Poe?s story, in the ways that both connect perception of external objects to the potential for intersubjective experience, dramaticaly underline the connection betwen these two tropes, seing and connections with other minds. Again and again the two are figured together, through triangles of joint atention. Finaly, a focus on 129 joint atention makes it clear that To the Lighthouse centers on a puzzle. ?Think of a table when you aren't there? has the same structure as a Zen koan or child?s riddle, one whose answer, I wil argue, depends not only on Ramsey?s philosophical debate over the relationship betwen existence and perception, but also on the crucial importance of shared experience. What does it mean to se something if no one is there to se it with you, or something that no one else can? A reading of the two texts centered on joint atention iluminates the common ground betwen these two apparently very diferent writers. Poe, for whom I claim the status of a theorist of consciousnes, and Woolf, the theorist of iredeemably internal subjective experience, both produce naratives not just of consciousnes but of interconsciousnes. In each, the internal and subjective are realy about that which is external, triangulated upon, and intersubjective. The works and the characters within slowly piece together evidence to study the interior lives of others, mysteries that may or may not be solved. The Modernists and the Detectives Where the modernist novel renounces linearity, logical order, and the stability of the outer surface of reality, the detective story embraces and valorizes them. D. A. Miler?s seminal study The Novel and the Police (1988), for example, presents twentieth-century detective fiction as the inheritor of a textual trend that blossomed in the ninetenth century, in which the representation of crime and policing cemented the novel?s role as an oblique form of social control. The conventions of the detective 130 story are soothing, orderly, and contained; they asume that the world is rational and knowable. Surveilance is justifiable and temporary. Disruptions wil be setled, transgresions wil be located and remedied, crimes wil be punished. In short, detective fiction caries water for the status quo. As such, the ?ordered surface of life and reality? and ?determination of an onerous plot? (Bradbury and McFarlane 1991:393) that modernism rejected are precisely where detective stories are most at home. And yet, as Gertrude Stein wrote in her 1935 esay ?What Are Master-Pieces and Why Are There So Few of Them?? one might also cal the detective novel ?the only realy modern novel form that has come into existence? (p. 2), and detective stories were certainly popular with many of the same readers who formed the primary audience for experimental modern literature. Some commentators (e.g. DiBatista 1996) explain this connection by arguing that detective fiction functioned primarily as an escape from what Marjorie Nicolson (1929) caled the ?crimes? of modernism, such as pesimism, uncertainty, and formlesnes. In this characterization, detective fiction served as an antithesis of and antidote for overdoses of modernist dis-ease. Others (McHale 1987; Irwin 1994) argue that the ritualization and formalism of the detective story and its epistemological preoccupations make it not modernism?s opponent, but its aesthetic aly. Indeed, a number of prominent Anglo-American modernist writers, most famously Gertrude Stein and T. S. Eliot, wrote seriously and approvingly on the subject of detective fiction. We might wel ask, then, whether it is more iluminating to consider the two kinds of texts together or apart; if together, in opposition or in alignment; and on just what lines it is productive to compare them. 131 A clasic reading of the modernist novel as exemplified by Woolf, put forward most influentialy in Auerbach (1946), is that it innovates by inverting the Victorian novel?s standard relationship betwen characters? internal states and the outside world. As this acount would have it, the clasic realist novel that reached its height in the ninetenth century took an interest in characters? motivations, thoughts, felings, and desires primarily insofar as they served as explanations of or triggers for events in their world. These texts are often at great pains to impres upon the reader how everyone in a given situation has reasons for their actions, but this is diferent from treating actions as primarily of interest for their efects on the internal landscape. Even a novel like Middlemarch, which invests its characters with very rich inner lives that form the bulk of the novel?s content and interest, the events and qualities of those interior lives are understood largely in terms of how they ?prepare and motivate significant external happenings? (p. 538). The modernist move is to turn this causal relationship on its head, creating situations in which external objects and events are of interest primarily for the results they create in the mind of a character. Atention to the mechanics of joint atention in ?Murders in the Rue Morgue? reveals that, in fact, Poe too?inhabitant of the ninetenth century though he might be?turned this relationship back to front. It?s easy to mis just how true this is, for several reasons. First, we are so used to thinking of this story primarily in its capacity as the mother of al detective stories that commonplaces about the genre adhere to it automaticaly; it is very dificult to step back and take it on its own merits. Furthermore, the events of the story are flashy and sensational, and they become the hook on which readers tend to hang their summaries of the tale: it?s the one where the 132 monkey did it! What?s more, it is common wisdom that crime stories are first and foremost about crimes, and motives are only of interest insofar as they contribute to the solution of the mystery. As Leonard Woolf wrote in 1927: It is a curious fact that a crime, a mystery, and a detective wil make almost any novel, however bad it may be in other respects, readable?. But the mystery?s the thing, and the writer who starts a detective story must stick to the mystery. He cannot expect us to stop and complacently admire him writing like Henry James or Mr. Compton Mackenzie for a dozen pages while the story is forgotten. (727) But on closer inspection, the real meat of ?Rue Morgue? is not the crime; that is only of note because of the ratiocinative proces it leads our detective to enact. The real subjects of the story are, first, an almost Joycean string of asociations running through the mind of the narator, and the way that Dupin is able to aces and share in those thoughts; and second, Dupin?s great chain of asociations prompted by his observations of the Rue Morgue and its inhabitants. In both cases, external events and objects ?serve to release and interpret inner events,? as Auerbach said of Woolf?s prose. And, as we wil se, in both To the Lighthouse and ?Murders in the Rue Morgue,? the relationship betwen the external and the internal is more sophisticated yet; the external and perceptible do not just serve as triggers for the events of a single consciousnes, but as a locus for the potential for intersubjective experience. New technologies of forensic science that first appeared in the ninetenth century served in detective fiction as a powerful model for acesing the minds of others through atention to the physical world. 8 Forensic investigation licensed a 8 For an excelent survey of the detective story and its connection to the development of both forensic science and the cultural authority with which it was invested, se Ronald R. Thomas' Detective Fiction and the Rise of Forensic Science (2000). The rise of criminology as a discipline in the ninetenth century also reflected a profound 133 special kind of interpretive strategy in which the detective could ?read? clues that were inscribed in the body and other physical traces of a suspect in order to reveal the solution to a mystery, including the motives and secret knowledge of the criminal mind. Freud, whose influence on Modernist literary fiction is wel known and dificult to overstate, similarly presented the mind as something acesible through interpretive practices applied to external manifestations of the self. Dreams, denials, and slips of the tongue could be subjected to a specialized technology of interpretation that would reveal depths of the mind hidden even to the mind?s own possesor. Both methodologies thus emphasized a distinction betwen physical surface and mental depths and proposed causal relationships betwen the two in which a skiled practitioner could use the surface as a key to unlock the mysteries beneath. The centrality of joint atention in communication means that items and events that are understood to be the subject of joint atention register as more salient, more memorable, and more important than other input. This suggests that there is a deep reason why external objects, properly contextualized, can serve so wel as a means of conveying information about internal states. I suggest that what we know about the organization of social cognition reveals that clasic detective fiction and modernist naratives that experiment with stream-of-consciousnes interior monologues converge on a single kind of conundrum, a ?puzzle of joint atention.? Both frequently present situations in which tracking a character?s gaze as it moves from new interest in the idea of the ?criminal mind? as a type, as wel as the idea that the individual minds of individual criminals were both of interest and amenable to (and in need of) scientific elucidation. 134 object to object in his or her environment serves as the key that unlocks the hidden secrets of that character?s inner thoughts. These puzles and their solutions represent the amplification of a very real behavior. In mystery stories, the detective is the possesor of this special aces to and heightened understanding of the significance of his felows? gaze. In the modernist novel of consciousnes, it is the reader who is invited, first and foremost, to take the part of the atentional detective. Stylistic experiments present readers with their own puzzles in which they have aces to a point of shared atention and are left to deduce or intuit the motivation behind the atending. Points of character, states of mind, and motivating transitions are to be filed in folowing the trail of the focalizing character?s atention. This puts the reader in the position of the detective Dupin: following a chain of referents, one to the next, like a path of stepping stones leading to an inevitable conclusion. Reading Minds in the ?Rue Morgue? Edgar Alan Poe?s ?The Murders in the Rue Morgue? (1841, reprinted in Poe 1992) is widely recognized as the very first modern detective story. While ?The Purloined Leter? has played a much more prominent role in the history of critical theory and poststructuralist discourse?providing a tantalizing metaphor for the relationship betwen language and the unconscious and serving as the jumping-off point for a particularly famous debate betwen Jacques Lacan and Jacques Derida? the earlier story is arguably more influential in the domain of literature itself. It 135 contains the first instances of many tropes that were destined to become standards of the genre: the detective whose ecentricities are matched only by his powers of reason; the friend and companion who records his uterances and narates the tale; the inept police whose actions throw the detective?s briliance into sharper relief; and, finaly, the detective?s astonishing chains of deduction which are explained step by step and invariably prove correct. Central to the detective mystique is the prerequisite that sound reasoning and solid facts back the detective?s conclusions, even if only the briliance of the detective is sufficient to recognize the facts and acomplish the reasoning. One of the best-known pasages in al of Poe?s fiction comes from the opening of this story. In this scene, Dupin demonstrates his signature gift as a detective, the fame of which Sherlock Holmes admits and disparages in A Study in Scarlet, describing it with scorn as a mere ?trick? to which he would never stoop. In any case, the narator explains that Dupin had a ?peculiar analytic ability?: to Dupin, ?most men? wore windows in their bosoms? (p. 144) that make their thoughts transparent to him. The narator then provides an example of this gift in action: The two men have been walking in companionable silence through the strets of Paris for some fiften minutes. Suddenly Dupin pierces the narator?s reverie with a remark: ?He is a very litle felow, that?s true, and would do beter for the Th??tre de Vari?t?s.? This comment is so perfectly calibrated with the narator?s train of thought that at first he simply agres. But soon enough he realizes that he should be astonished, and duly expreses his amazement. He had indeed been thinking of an actor named Chantily 136 and furthermore, as Dupin helpfully supplies, been thinking ?that his diminutive figure unfited him for tragedy? (145). How, the narator marvels, was Dupin able to read his mind so uncannily and acurately? Dupin is happy to explain the analytic method by which he penetrated the narator?s thoughts. He reminds his friend of the fruiterer who ran up against him as they first entered the stret, fiften minutes before, knocking the narator against a pile of paving-stones and causing him to slip and strain his ankle. This collision clearly set in action a train of asociations, says Dupin, which he was able to reproduce through close observation of the narator?s movements and a series of deductions based on what he knows of his friend?s interests and paterns of thought. To the narator, perhaps the most astonishing aspect of this deductive feat is the ?apparently ilimitable distance and incoherence betwen the starting point and the goal? (146) that any chain of asociative thought evidences upon reflection, even to oneself. How much more astonishing, then, that the detective could follow the thread from outside, as he did! Of course this feat of mind reading is fantastic, a point which historicaly has irked many commentators. R. Austin Freman (1924, reprinted in Haycraft 1976:16- 17), for example, complained: What claims to be a demonstration turns out to be a mere specious atempt to persuade the reader that the inexplicable has been explained; that the fortunate gueses of an inspired investigator are examples of genuine reasoning. A typical instance of this kind of anti-climax occurs in Poe?s ?Murders in the Rue Morgue? when Dupin follows the unspoken thoughts of his companion and joins in at the appropriate moment. The reader is astonished and marvels how such an apparently impossible feat could have been performed. Then Dupin explains; but his explanation is totaly unconvincing, and the impossibility remains. 137 Yet however incredible the details of Dupin?s methods may be, they are far from entirely divorced from the plausible, indeed commonplace, methods of ?mind reading? that people use every day. Instead, they represent a kind of embelishment of these natural cognitive strategies, amplified to a preternaturaly efective degre. The physical observations on which Dupin hangs his implausible inferential leaps turn out to focus almost entirely on the direction of his friend?s gaze. In imediate reaction to the collision with the fruiterer, Dupin explains, the narator ?turned to look at the pile,? making it clear to Dupin that the paving stones were at the forefront of his companion?s thoughts. As he walked away from the site of the collision, he ?kept [his] eyes upon the ground - glancing, with a petulant expresion, at the holes and ruts in the pavement.? By this, Dupin explains, he could se that the narator ?was stil thinking of the stones? (146). Next, the narator brightens upon seing the diferent, experimental paving style of the Aley Lamartine. Dupin ses his friend?s lips move, and infers correctly that he is making reference to the new pavement he is looking at so approvingly. The basis for the detective?s inferences then shifts momentarily away from his friend?s eye movements to his own background knowledge about what kind of person his friend is and what ideas are likely to be prominent in his memory. From this foundation, Dupin can surmise that the particular word his friend would use to refer to the pavement is ?stereotomy?. Knowing his friend as he does, and remembering conversations they've had in the recent past, he can deduce further that ?otomy? would cal to mind ?atomy?; atomies would cal to mind the philosopher Epicurus; and the thought of Epicurus would cal to mind their recent conversation about how 138 litle support his theories have found in contemporary astronomy. Finaly, having made this independent chain of deduction, Dupin is now ready to test it with a fresh empirical observation. Once again the object of his observations is the narator?s gaze: ?I felt that you could not avoid casting your eyes upward to the great nebula in Orion, and I certainly expected that you would do so. You did look up; and I was now asured that I had correctly folowed your steps..? (146). Dupin?s pyrotechnic display of asociative logic, then, demonstrates not just any preternatural talent for mindreading?he is specificaly a master of tracking and making sense of the direction of his companion?s gaze. For Dupin, shifts in his companion?s gaze through the landscape of objects that surround him are the symptoms by which his inner thoughts may be diagnosed, the clues to a puzzle which can be solved and whose solution can be explained to both the reader and to others within the world of the text. He is master of a model of social cognition based in triangulating atention to objects in the world. In the terminology of social cognition and joint atention, gaze-following refers to the act of monitoring the direction of another person?s line of regard and using this information as a basis for making asumptions about that person?s mental state. Remember that the ability to engage in scenes of joint atention is evidenced in a suite of several diferent new interactional skils that emerge over the course of several months in early childhood, al of which demonstrate a newfound ability to ?tune in? to other people?s atitudes toward entities in the outside world. Gaze- following is one of the earliest of these skils to appear and provides important scafolding for language acquisition in general and lexical acquisition in particular 139 (se, e.g. Bruner 1983; Baldwin 1991, 1993a). The ability to monitor the visual atention of others in this sophisticated way is not al there is to the social cognition and cultural learning upon which humans' unique skils in language rest. It is, however, a very wel documented and powerful tool for coordinating understanding that is central to our social and linguistic development from very early in life. Al of this is to say that the demonstration of Dupin?s ratiocinative prowes in ?Rue Morgue? is intimately connected with some of our most basic means of generating meanings when presented with ambiguous or unlearned signs. Critics, following the lead of Lacan?s seminal analysis in ?Seminar on ?The Purloined Leter?? (1966, reprinted in Muller and Richardson 1988), have historicaly turned to that tale as Poe?s go-to alegory for the semiotic. But despite the central place that ?Purloined Leter? holds for psychoanalytic and postfoundational theories of the sign, I would argue that ?Rue Morgue? in many ways presents an even beter parable for the semiotic experience. It resonates both with a contemporary cognitivist understanding of the relationship betwen language and the mind, and with a characteristicaly modernist approach to linguistic representation and the knowability of other minds. Central to Dupin?s analysis in the titular case is his insight into a striking feature of the depositions collected by the police: no one can agre on the question of what language the murderer spoke. Several witneses have testified that on the night of the murder, they heard two voices ?in contention? inside the house. One is gruff and can clearly be heard to say several things in French, including ?mon Dieu.? Every one of these witneses asumes that the other voice also belongs to a human. 140 However, each one believes that this second person is speaking in a diferent language from any of those suggested by his felow itneses. The only point in common is that the language is both foreign and incomprehensible to the witnes in question. Dupin runs through the list (p. 155-6): One Frenchman believes that the voice belonged to a Spaniard and says that he ?might have distinguished some words? if he only knew any Spanish. Another was ?convinced by the intonation? that the language was Italian, a language, again, that he did not speak. A Dutch witnes, testifying through an interpreter because he knows no French himself, claims that both voices were speaking French. An English tailor ?does not understand German,? but thinks that the voice may have been speaking in that language. Finaly, an Italian confectioner hypothesizes that it was Russian, though he has ?never conversed with a native of Russia.? Here, then, we se a Paris house transformed into a miniature tower of Babel: a modern jumble of mutualy impenetrable signs emanating from a peculiarly modern jumble of multinational Parisian merchants. This turn of events points strikingly to a postcolonial reading in which language becomes cacophony, underlining anxiety about the encroachment of the foreign. Indeed, the revelation that the voice belongs not to a human, but to an orangutan, serves in no smal part to link together, rather than diferentiate, the foreign and the inhuman. Dupin wryly mentions, ?You wil say that it might have been the voice of an Asiatic?of an African. Neither Asiatics nor Africans abound in Paris? (p. 156). The one exception, apparently, can be found in the person of the bloodthirsty ape whose Asiatic, if non-linguistic, voice is so central 141 to the tale. In this way, the culprit?s animal nature reflects not only on his species but also on his foreignnes. 9 Beyond these obvious identity-related implications of the linguistic confusion of the tale, the semiotics of this scene also demonstrate (in fact, take for granted) an even more deeply rooted impulse: the human inclination to look for communicative intentions and minds like their own wherever they can find them. Almost everyone in this litle tale over-atributes intentions, particularly communicative intentions, to what turns out to be a creature who doesn't have them. These exceses serve to highlight how much of the semiotic experience rests on grasping at relative straws of behavioral evidence of other people?s mental states and communicative intentions. In 9 Here, of course, I pas over many other relevant historical readings of ?Rue Morgue?. One that deserves at least brief mention is Elise Lemire?s (2001) discussion of the many ways in which the orangutan?s depiction serves as a titilating exploitation of the acute racial anxieties and related violence that formed the background context for Poe and his readers in 1830s America. By choosing an orangutan as his kiler and by linking the primate?s non-speech cries with the sound of an ?African? voice, Poe taps into a set of wel-established cultural tropes of his era (realized and reinforced in museum displays, cartoons, and commonplaces) in which black people were supposed to closely resemble simians in general and, often, ?orang-outangs? in particular. Lemire points out, further, that by arming the Rue Morgue?s orangutan with a barber?s razor, Poe was making this mapping inescapable for his American readers. Barbering was considered the signal profesion of entrepreneurial fre black men of the period, an asociation that neatly tied together anxieties over cultural and economic incursions into white society with more visceral, sensational fantasies of white vulnerability, in the form of the client?s bared neck and the barber?s shining blade. Thus in its depiction of a bestial escaped kiler who invades the bedchamber of white women and slits their vulnerable throats, the story serves at least in part as a racialy marked cautionary tale: a reactionary, sexualy charged indulgence for readers who were inclined to make these connections. At the same time, this alegorical mapping can also be read as a critique of white Southerners? inability to recognize anger, agency, and the potential for rebelion in the African-Americans they enslaved, as the Maltese sailor underestimates and mistreats his simian charge. Poe?s complicated political and rhetorical position on these maters thus is also an exercise in alternately exposing and indulging various problematic shared ideological perspectives: another kind of puzzle of joint atention. 142 ?Rue Morgue,? the mark of the truly briliant man is the ability to sort out what is and what is not a real sign of those intentions. The leson is not that one should avoid this kind of thinking?quite the contrary, as per Dupin?s virtuoso performance on the cobblestones. Rather, the moral is that you should do it wel. Virginia Woolf would later imagine an exactly similar mechanism for her characters? atempts to apprehend one another?s minds, organized around triangulated gaze and the sense that shared aces to a single object of visual atention could or should serve as a crucial site of connection betwen consciousneses. Where in Woolf this mechanism serves as a locus of the dificulty of sharing subjective experience, however, Dupin epitomizes the pinnacle of ?efortles? mind reading that appears so commonly in the literature on intersubjectivity in modern cognitive science. The scientific study of how communicators understand one another wil naturaly tend to take the asumption that they usualy do understand one another as given. In particular, I am thinking of the common notion in procesing and psychological acounts of language use that, while misunderstandings and erors are easy to notice, a primary question for researchers should be to explain how we manage to get so much right with so litle apparent efort. 10 In this emphasis on ease, then, Poe is on the one hand extraordinarily up-to- date and on the other the product of a characteristicaly mid-ninetenth-century confidence in knowability, sharability, and the rational. Dificulties of communication appear as the results of either social obstacles or individual intelectual shortcomings 10 Se, for example, Garod and Pickering (2004), ?Why Is Conversation So Easy?? For a contrary approach, se Taylor (1992), Mutual Misunderstanding: Scepticism and the Theorizing of Language and Interpretation. 143 (?without educated thought,? Dupin claims, even the most clever investigators wil er ?continualy?), not as structural features of consciousnes or language. The products of individual consciousneses can always be measured against and deduced from a shared physical reality. And so the semiotic logic of ?Rue Morgue? depends upon divining atentions and intentions from the kind of physical symptoms of mental states we se in canonical scenes of joint atention. This is not only true for the scene on the cobblestones; it also resonates through the central mystery and its resolution, in which it is revealed that what was asumed to be a pair of murders were in fact the acts of a ferocious escaped orangutan. The semantics of ?murder? both as a legal term and in its everyday folk usage are complex (se, for example, atempts in Maley 1985 and Langford 2000 to provide exhaustive analyses of the word and its meaning in use). Chief among the defining properties, however, in any analysis, are the intentions and intentional capacities of the perpetrator. One cannot ordinarily be murdered by a faling tre, a railway acident, or a charging bull. Thus the title of the story itself is a play on the relationship betwen meaning and intention. There were no murders in the Rue Morgue, not because the result for the victims was any diferent than if they had been murdered, but because the correct reading of the events reveals that the agent of their demise could not have the kinds of intentions necesary to make their deaths count as murders. Again, Dupin?s great feat is to take a physical trace of another being?s focus of atention and use it not just to form a notion of what that agent did, but what he thought, or did not think. What?s more, in the world of the text, he is not remotely alone in this view. His peculiarity is 144 in his succes, not in his approach. No les than Woolf, Poe is a theorist of consciousnes, what it consists of, and how it can be apprehended. The consciousnes at the heart of his theory is a shared one. And Woolf no les than Poe, we shal se, hangs her acount of the intersubjective on triangulation at a distance through the concrete, perceptible objects of the physical world. In To the Lighthouse Shared Consciousnes and Joint Atention None of the characters in Virginia Woolf?s To the Lighthouse (1927, Harvest edition 1981) shares Dupin?s unering gift. His talent for gaze-following as a tool for diagnosing consciousnes is, as we have sen, not only far-reaching but preternaturaly reliable. None of Woolf?s characters are so fortunate. Yet To the Lighthouse is also structured in many crucial ways around the way that visual atention serves as a key physical manifestation of characters? mental states. As in ?Rue Morgue,? the connection betwen thought and sight requires triangulation through external objects; this is not the notion that ?eyes are the window to the soul?. Sharing atention to objects in the outside world is critical for Woolf?s characters as an opportunity?even if a frequently fraught or thwarted opportunity?for closenes and sympathy. Characters repeatedly find themselves looking at the same object and experiencing a frison of connection; they find themselves wondering about the degre to which they are transparent, or to which they can inhabit the thoughts of another. 145 This treatment of triangulated gaze, then, is a point of continuity betwen Woolf?s modernism and Poe?s romantic materialism, as wel as a break from the major models of intersubjectivity found in clasic realist fiction. The Victorian social novel, like high modernism, had no shortage of characters with dificulty understanding one another?s minds, particularly their motivations and felings. Those failures of sympathy, however, were typicaly rooted in confining social structures and individual stubbornnes rather than the ontological division betwen the self and the external world that dominates To the Lighthouse. Under the correct circumstances, the correct people could overcome these obstacles through the combined application of humblenes, generosity, and physical contact. (Compare, for example, the climactic consummation of felowship betwen Dorothea and Rosamond in Middlemarch, in which both emotional and intelectual understanding are precipitated by a physical embrace.) For Woolf, it is in the act of looking that the self can most closely approach nakednes. One may on occasion find that the ?shel-like covering which our souls have excreted to house themselves, to make for themselves a shape distinct from others, is broken, and there is left of al these wrinkles and roughneses a central oyster of perceptivenes, an enormous eye? (1942:22). The possibility of joint atention is thus also a source of anxiety, as when Lily tries to avoid looking at her painting while Mr. Ramsey is watching: ?But so long as he kept like that, waving, shouting, she was safe; he would not stand stil and look at her picture. And that was what Lily Briscoe could not have endured? (TL 17). 146 One interpretation of this moment is that she simply does not want Ramsey to se her painting and what the painting itself exposes of her self, and that she fels that that his having sen it wil somehow change or ruin her idea of what the painting is. Al this is certainly correct. But there is more to the scene than that. Importantly, Lily does not want to se him se the painting, or fel him se it, lest she be forced to acknowledge or experience the force of his thoughts. This scene has its partner in the third part of the book, after Mrs. Ramsey?s death. Mr. Ramsey and Lily Briscoe stil bridle at the thought of being forced, through the confluence of their gaze, into a confluence of the minds (p. 151). An ?awful pause? follows Lily?s resistance to making any expresion of sympathy (though she is is al too aware, and correctly, that this is what he wants). Both look at the sea. ?Why,? thinks Mr. Ramsey resentfully, ?should she look at the sea when I am here?? He wants her to understand his thoughts and desires; he does not want to be impinged upon by hers. Lily, in her turn, has already been thus imposed upon. His ?enormous flood of grief? and ?insatiable hunger for sympathy? are al too palpable. It is the thought of just such a moment of uncomfortable communion, made worse and more personal by the idea that her painting would be at the apex, that the younger Lily fears the most. To be confronted with Mr. Ramsey in such a way is the most alarming prospect, but the prospect of this kind of intimacy with almost anyone is too much to bear?if ?Mr. Tansley, Paul Rayley, Minta Doyle, or practicaly anybody else? should come up behind her and look at it, she thinks, she would have to turn the canvas face-down on the gras rather than suffer through such a charged and intimate experience. But when Wiliam Banks walks up, she lets it be; and thus 147 the simple sentence ?Wiliam Bankes stood beside her? (p. 17) conveys a quiet and touching empathy. Indeed, most of the moments in which people do succesfully understand one another in the novel can be found in these scenes of joint visual atention betwen Lily Briscoe and Wiliam Bankes. Some are described so that we know for certain that these moments of shared seing are also moments of real mutual understanding, as when they walk to look together at the sea. ?They both felt a common hilarity, excited by the moving waves; and then by the swift cutting race of a sailing boat? and then, with a natural instinct to complete the picture, after this swift movement, both of them looked at the dunes far away, and instead of meriment felt come over them some sadnes? (p. 20). And because this novel is not populated with genius detectives but with Woolfian modernist subjects, Lily and Wiliam?s natural shared inclination to look at the dunes after the boat is not a Dupin-like parlor trick but instead a manifestation of their individual and unusual degre of natural commonality. Other times the understanding that results from these occasions of joint atending is not explained, and the shared atention is instead simply presented as a signifier of a fleting but real connection betwen two minds. These moments often appear in places of poetic emphasis, so that they are lent added weight. For example, chapter four of the first section ends with a breathles 153-word sentence that conveys something of Mr. Ramsey?s hectic but frozen response to having been discovered in a moment when he thought we was alone. After over a hundred words, the hiatus ends: ??he turned abruptly, slamed his private door on them; and, Lily 148 Briscoe and Mr. Bankes, looking uneasily up into the sky, observed that the flock of starlings which Jasper had routed with his gun had setled on the tops of the elm tres? (p. 25). The result is that we come to rest on this shared seing; it stands against the vast and unsympathetic distance in their encounter with Mr. Ramsey. Its shared nature is uncontroversial?we are simply told that the two observe together. Note also that the succes of this act of joint atending is in no way the result of any active or deliberate atempt to penetrate a distant consciousnes. The Reader as Dupin While Woolf?s characters do, like Dupin, pay conscious atention to the atentions of their felow characters, and strive to gain some understanding of one another?s inner thoughts thereby, the succes of these endeavors for the characters themselves is limited at best. Instead, it is the reader who is the true Dupin of the novels, the reader alone who can appreciate the full sequence of ideas prompted or released by the external events that the characters se. But this experience of course is not true joint atention either. Joint atention must be, by definition, mutualy manifest, and no character wil ever appreciate that the reader is sharing her perspective. On the one hand, the lopsidednes of this relationship is part of what makes it a strong paralel to the version of this dynamic that we saw in Dupin; it is what makes the reader a detective, and what makes these referential triangles puzzles rather than otherwise. But it also both underlines and breaks open the lonelines at the heart of the novel. These experiments in form are Woolf?s atempt to solve the problem she described in Mr Bennet and Mrs Brown (1924:18) of how to impart ?that vision to 149 which I cling? and ?the appaling efort of saying what I meant?. The succesful depiction of these characters? lonelines and the failures of their atempts to share with one another the visions to which they cling, then, is a liberation, because it is a connection with what sems to be a real, or at least a convincing, other mind. And yet under the best of circumstances, the understanding goes only one way. The detective is a lonely figure too. The disparity betwen the reader?s ability to close the intersubjective triangle and that of Woolf?s characters is ilustrated at length in the dinner party near the end of ?The Window,? in which Mrs. Ramsey devotes her considerable artistic talents for social choreography to al the members of her family and house guests at once, in a single gathering. In this sequence, the characters are simultaneously coordinated collaborators in the single achievement of a social event??they were al conscious of making a party together in a hollow, on an island? (p. 97)?and absorbed in their own disjunct individual projects. Lily, for example, devotes herself both to salvaging social disharmonies on Mrs. Ramsey?s behalf and to thoughts of her own half- finished painting. Mr. Tansley dedicates himself to the project of bolstering his own sense of importance and distance from the Ramseys, in defiance of his own sense of social inferiority. Mrs. Ramsey, meanwhile, takes a keen interest in where the people around her are looking, and makes a number of hypotheses about their states of mind based on these observations. ?There was Rose gazing at her father,? she notes, ?there was Roger gazing at his father; both would be off in spasms of laughter in another second, she knew..? (p. 96). She similarly observes her daughter Prue?s atention to Minta: 150 ?She kept looking at Minta, shyly, yet curiously, so that Mrs. Ramsey looked from one to the other and said, speaking to Prue in her own mind, You wil be as happy as she is one of these days? (p.109). But these insights are unconfirmed and partial or fleting. Soon enough, Mrs. Ramsey is wondering what ?joke of their own? her children might be harboring: ?What was it, she wondered, sadly rather, for it semed to her that they would laugh when she was not there. There was al that hoarded behind those rather set, stil, mask-like faces..? (p. 109). Betwen these moments, Mrs. Ramsey notices that Augustus Carmichael and she have both been looking at the plate of fruit that Rose has aranged. Again she fels that the confluence of visual atention brings the lookers closer together: ..to her pleasure (for it brought them into sympathy momentarily) she saw that Augustus too feasted his eyes on the same plate of fruit, plunged in, broke off a bloom there, a tasel here, and returned, after feasting, to his hive. That was his way of looking, diferent from hers. But looking together united them. (p. 97) But are they truly united, when he is clearly oblivious to her sympathetic felings? Or is Lily more nearly right when she thinks, of Mr. Bankes: ?She would never know him. He would never know her. Human relations were al like that..? (p. 92)? Al of these false starts and thwarted atempts at coordination mean that the reader alone inhabits the privileged position of the succesful detective, for only the reader is truly privy to the ?inner movements? of the Ramseys and their guests. The flow of information for the reader in this pasage is organized in much the same way as his companion?s thoughts are for Dupin. We know, for example, that when Lily ?catch[es] sight of the salt celar? on a particular spot in the patern of the tablecloth, she is reminded of her inspiration to move the tre in her painting, and of al her 151 asociated ideas about how her art ?saved? her from the ?dilution? of mariage (p. 102), but her felow characters do not. Gaze, again, is the organizing structure; the thread of the narative is the shifting atention of the characters as their gaze setles on this or that element of their surroundings. In this way, in addition to the characters' atention to the atentions of one another, we trace their mutual and solitary movements of atention with respect to objects in the room. When characters' atention drifts away from the conversation, their thoughts stil remain grounded in the visual apprehension of their physical surroundings. The focus of the text stil follows the focus of their drifting gaze. ?So they argued about politics, and Lily looked at the leaf on the tablecloth? (p. 94)? thinking, again, as she and the reader alone know, of her painting. Mrs. Ramsey ?looked at the window in which the candle flames burnt brighter now that the panes were black, and looking at that outside the voices came to her very strangely.. for she did not listen to the words? (p. 110). The appearance of a single object can also kick off a sequence of asociations that crosses from one character to another, as when the maid brings in a ?huge brown pot? of boeuf en daube. The sight of it leads Mrs. Ramsey to think of the thre days the cook had spent making the dish; of how she must be careful to choose a particularly nice piece for Wiliam Bankes; of how such a lovely dish wil ?celebrate the occasion,? prompting ?a curious sense? to arise within her, at once freakish and tender, of celebrating a festival, as if two emotions were caled up in her, one profound - for what could be more serious than the love of man for woman, what more commanding, more impresive, bearing in its bosom the seds of death; at the same time these lovers, these people entering into ilusion glitering eyed, must be danced round with mockery, decorated with garlands. (pp. 99-100) 152 Meanwhile, for some uncertain majority of the diners, we se, the candles serve as a distraction from Mr. Ramsey?s iritation over Augustus Carmichael?s second plate of soup; the candles draw atention to the plate of fruit, to the reflections in the window, and to the faces around the table. Mrs. Ramsey looks at the bowl of fruit and notices Augustus; Augustus looks at the fruit and notices something to eat. The same fruit prompts Mrs. Ramsey to embark upon a more extended chain of asociations: ?of a trophy fetched from the bottom of the sea, of Neptune?s banquet, of the bunch that hangs with vine leaves over the shoulder of Bachus (in some picture), among the leopard skins and the torches lolloping red and gold?? (p. 97). This kind of pasage has clear paralels in other experimental novels of the period. The ?Calypso? chapter of Ulyses, for example, introduces the reader to Leopold Bloom, painting a compeling and intimate portrait of Bloom?s internal landscape primarily through his chain of physical atentions and the asociations to which they give rise: ?He peeped quickly inside the leather headband? (p. 46); ?He crossed to the bright side? (p. 46); ?He approached Lary O?Rourke?s? (p. 47); ?He creased out the leter at his side? (p. 53); ?In the bright light, lightened and cooled in limb, he eyed carefully his black trousers? (p. 57); and so on. Each of these narated actions serves as the jumping-off point for a meditation prompted by a shift in Bloom?s atention to some new object. They are folowed always by a bare mention of some physical object, which we can take to be the current object of Bloom?s atention, often but not always physicaly present??Another slice of bread and butter? or ?White slip of paper??that sends Bloom and the reader together through a chain of asociations, from an advertisement trumpeting a German company?s plan to 153 plant eucalyptus groves in Palestine to thoughts of oranges, melonfields, and olive tres in in Jafa through memories of Molly?s tasting olives for the first time; from the thought of tisue-wrapped oranges in crates he moves by asociation to citrons, and by way of a pun to friends of his from an earlier time named Citron and Mastiansky, and so forth. Like Dupin tracing a glance at the cobblestones to his friend?s thoughts on Orion and the actor Chantily, the reader can se the connections betwen Bloom?s glance at a newspaper and his memories of old friends, and betwen Mrs. Ramsey?s sight of a plater of grapes and her thoughts of the god Neptune. In its emotional valences, the detective stories to which Ulyses, if not To the Lighthouse, has the closest afinity are not the Dupin tales but the works of Arthur Conan Doyle. This connection was first advanced in an early article by Hugh Kenner, ?Baker Stret to Ecles Stret? (1949, revised in Kenner 1956), which argues that the relationship betwen Stephen Dedalus and Leopold Bloom a recapitulation and elevation of the relationship betwen Sherlock Holmes and John Watson. ?Holmes and Watson,? Kenner observes, ?epitomize humanity disected into ratiocinative violence and sentimental virtue, the later avid of absorption into the former? (p. 170). So it goes with Stephen and Bloom. But in its methods I have highlighted here, Ulyses is more like Woolf and, perforce, more like Poe. Dupin?s cobblestones are the first link in a chain of supposition, the ancestors of Bloom?s butcher-shop kidneys and pasing clouds, as wel as of Lily?s ocean waves and kitchen tables. This brings us back to the kitchen table in question, which I argue can be read in a new and interesting way in light of social cognition, joint atention, and narative methods that thematize and capitalize on their mechanics. 154 A Kitchen Table When Only You Are There Early in the novel, Andrew Ramsey tries to explain his father?s work on ?subject, object, and the nature of reality? to Lily Briscoe: ??Think of a kitchen table then,? he told her, ?when you're not there?? (p. 23). Mr. Ramsey is a philosopher whose lofty intelectual thoughts prove of litle use to him in appreciating the human events in the house around him; but Lily, who is much more sensitive both to al the other minds around her and to the degre to which they are al isolated from one another, seizes on this image. It haunts her; it becomes a source for her art; in sum, it becomes something much more rich and compeling, in the framework of the novel, than Mr. Ramsey would ever know. The standard reading of this image focuses on absence and domesticity. Mary Jacobus (1986), for example, ties these together in a feminist psychoanalytic framework, linking Melanie Klein?s revision of the Freudian notion of ?object loss? with a post-Oedipal narative of gender relations. The kitchen table works in part as a figure of the things that are ?not there,? which echo poignantly through the book. First the lighthouse and then, after her death, Mrs. Ramsey, haunt the thoughts of every character, each one invested with al the more significance the les acesible it becomes. More imediately, in Andrew?s phrase Mr. Ramsey?s insistently masculine philosophy, in contrast to which ?the folly of women?s minds enraged him? (p. 31), is translated and transformed into something domestic, quotidian, and feminine. Kitchen tables are the domain of the Mrs. Ramseys of this world, not the philosophers. I would go a step further to argue that the kitchen table presents a transformation of the solipsistic Ramseyan philosophy into something new and 155 diferent, constructed out of the mechanisms of joint atention, in which experiences gain meaning in being shared. The work of Ramsey?s that Andrew is trying to describe to Lily with this phrase appears to be an entry in the epistemological and ontological debate over the role of perception and the importance of the observer for the nature of both knowledge and reality. Woolf?s contemporaries G. E. Moore and Bertrand Russel were hard at work on these very questions at the time To the Lighthouse was being writen. Moore and Russel were largely working in response to the Idealist philosophy of George Berkeley, articulated in his 1710 Treatise Concerning the Principles of Human Knowledge. In Berkeley?s formulation, al qualities of objects are no more and no les than sensory data, existing only so long as a sensate being perceives them. Color exists only through being sen, heavines only through being hefted, and so on. The new Realist philosophy espoused by Moore and Russel shifted the emphasis to perceptibility, rather than perceiving; that which exists is amenable to perception. The objects of knowledge then become neither sensations nor sensible objects, but logical constructions built out of them. Woolf herself was aware of and interested in these arguments. As Leonard Woolf wrote, ?Through us and through Principa Ethica the four others, Vanesa and Virginia, Clive and Duncan, were deeply afected by the astringent influence of [G. E.] Moore.. The colour of our minds and thought had been given to us by the climate of Cambridge and Moore?s philosophy? (1964:26). Russel too was a friend and influence, mentioned often in Woolf?s diaries and leters. 1 There is litle doubt that 1 Se J. K. Johnstone (1954), The Bloomsbury Group, and Harvena Richter (1970), Virginia Woolf: The Inward Voyage, for the seminal discussions of the influence of G. E. Moore?s Principia Ethica and Philosophical Studies, respectively. For the role 156 this philosophy afected Woolf?s writing, though there is some debate over precisely how and to what degre. Andrew McNeilie (2000), for example, argues that while Moore may have been important to the Bloomsbury group in general, he is far too prosaic to be considered a real influence on Woolf?s work. Ann Bansfield (2000:47), meanwhile, has argued that Woolf?s fiction manifests a merger of the old and the new ontologies, adopting, as she puts it, ?the new philosophical realism inoculated with Berkeley?s Idealism?. I would suggest that what is most important for Woolf?s aesthetic is not so much any one of these metaphysical positions as the elements around which the debate centers, and particularly the idea that there is a problem, a puzzle or enigma, surrounding the relationship betwen what is, what can be known, and what is perceived. This puzzle is more important than any of its solutions. Further, now that we are armed with an understanding of the significance of triangulated atention and its implications for intersubjective thought, a close reading of Lily?s ?kitchen table? suggests a whole new set of asociations in which it does not stand for the role of perception in being, but instead raises the possibility of a radical claim for the role of interpersonal perception and understanding for epistemological concerns. As is manifestly demonstrated by the many scenes above in which Woolf clearly hangs the elusive possibility of communication betwen our separate mental worlds on acts of mutualy manifest shared atention to objects in the world, in To the Lighthouse, to se an object is also crucialy to raise the possibility of genuine connection with another person. Not to se it is to remain alone. of Bertrand Russel?s philosophy in Woolf, se Ann Banfield (2000), The Phantom Table. 157 And so it should be unsurprising, in light of this reading, that Lily ?ses? the table not when she is trying to apprehend the nature of knowledge or existence, but through and because of her atempt to appreciate and sympathize with what she imagines Mr. Ramsey?s very diferent mental life must be like: now she always saw, when she thought of Mr. Ramsey?s work, a scrubbed kitchen table. It lodged now in the fork of a pear tre, for they had reached the orchard. And with a painful efort of concentration, she focused her mind, not upon the silver-bossed bark of the tre, or upon its fish-shaped leaves, but upon a phantom kitchen table, one of those scrubbed board tables, grained and knotted, whose virtue sems to have been laid bare by years of muscular integrity, which stuck there, its four legs in air. Naturaly, if one?s days were pased in this seing of angular esences, this reducing of lovely evenings, with al their flamingo clouds and blue and silver to a white deal four-legged table (and it was a mark of the finest minds to do so), naturaly one could not be judged like an ordinary person. (p. 23, italics added) There are several things worth noticing about this pasage. First, Lily has completely abandoned?if she ever possesed?any thought of the original philosophical puzzle for which this kitchen table putatively stands. This means on the one hand that her atempt at understanding Ramsey is necesarily doomed from the start; on the other, it opens up the possibility that Andrew?s expresion of the puzzle may serve to raise a new and perhaps superior, or at least more interesting, line of philosophical thought. Next, I would note that, despite this fundamental departure from Ramsey?s starting point, Lily is indeed atempting to construct a moment of sympathy with another person, and again this is figured as an atempt to se the same thing as that person. This time the atempt to gain understanding via a shared object of visual atention is an even more ambitious task, as the potential shared object of atention is abstracted away from being an easily acesible shared basis for common ground 158 betwen Lily and Ramsey in a number of ways. The kitchen table is imaginary, not real. The phrase itself is Andrew?s, not Ramsey?s. Lily should know (if she thinks about it) that Ramsey is probably not realy thinking of a kitchen table when he thinks about his work; his work is not about any kitchen table, much les a particular one. And yet the more intently she tries to capture for herself some sense of what it is like to think like him, the more vividly she imagines her phantom table. She elides, perhaps acidentaly, perhaps involuntarily, perhaps just metaphoricaly, the original notion of imagining the general tenor of Ramsey?s thoughts with the particulars of ?this seing??this ?reducing of lovely evenings, with al their flamingo clouds and blue and silver to a white deal four-legged table,? which is, in fact, a product only of her own vivid visual imagination, and in its every detail a sign, moreover, of the kind of thinker that she is and that Ramsey is not. Indeed, there is no reduction of an evening to a table; for Ramsey, such a thorough and thoughtful contemplation of the table itself is not possible, and for Lily it is impossible to forget or ignore the beauty of the evening and its ?flamingo clouds.? What we have in the image of the table is thus an impresive condensation of the double bind surrounding the communicability of internal consciousnes that stands at the heart of the Woolf project as I se it. The intersubjective triangle of joint atention serves throughout the novel as a means by which characters can, however briefly, have real insight into one another?s mental lives. But more often, characters who can se that potential are disappointed in practice; they fail to maintain a real understanding, or they find that when the intentions and desires of another person do become obvious, it is as much an imposition as an insight. Yet the novel itself and its 159 methods turn to the same mechanisms in a stil-hopeful atempt to make the same connections work for the reader, if not for the characters being read about. The table is both a signifier of this intense drive to sek connection and a signal example of its failure. At the same time, it provides a signal succes in the vivid portrayal of Lily?s consciousnes that it provides for the reader. And once again, the mechanism in play is (an atempt at) triangulated visual atention. What?s more, the vigor and vividnes with which Lily makes her atempt is salutary in itself?it does her good and speaks wel of her that she should think this way. Thus, Lily?s encounter with the kitchen table serves to transform the Russel-Moore-Ramsey question of the relationship betwen existence and abstractly defined perception into an asertion about the relationship betwen shared perception and shared existence, a crucial distinction that would be les visible without an understanding of joint atention. Sumary As a theory ought to do, the framework of literary joint atention ofers a genuinely new ay of looking at individual texts. It alows us to understand the putatively internal and subjective elements in both Poe and Woolf as instead focused on the external, the triangulated upon, and the intersubjective. These readings suggest that stream of consciousnes is not primarily a device for geting inside some subjective experience, but a method for and depiction of sharing experience, in which consciousnes itself is something that exists not inside a person but in a physicaly 160 grounded intersubjective space. It is the quality of this shared consciousnes that gives our consciousnes our shape. These observations invite several novel readings of these texts, both separately and together: 1. Atention to the mechanics of joint atention in ?Murders in the Rue Morgue? reveals that Poe anticipates modernist experimentation in his use of external objects and events as triggers for internal states. 2. Even more interesting, it turns out that the internal states in question in both cases are intersubjective in nature, and that they are not incidentaly but criterialy formulated by ascribing special significance to shifts in visual orientation to the outside world. 3. In ?Rue Morgue,? Dupin epitomizes the pinnacle of ?efortles? mind reading familiar from the standard treatment of intersubjectivity in modern cognitive science. On the one hand, this makes the story a particularly compeling parable of the semiotic experience in a cognitivist age; on the other, this quality is a legacy of its own era. This ilustration of certain points of congruence betwen contemporary scientific treatments of communication and a ninetenth-century genre asociated with social control and philosophical na?veit? may incidentaly provide a bit of insight into why many humanists fel antsy about the prospect of bringing work in cognitive science to bear on traditional humanist subjects like literature or history. 161 4. Woolf, by contrast, concentrates on dificulty. But the central tropes and mechanisms of triangulated visual atention are the same. 5. This treatment of triangulated gaze, then, is both ? a point of continuity betwen Woolf?s modernism and Poe?s romantic materialism; and ? a discontinuity, for both Poe and Woolf, with some of the prevailing models of intersubjectivity found in clasic realist fiction. 6. The framework of literary joint atention also yields a new reading of the ?kitchen table? of To the Lighthouse. It now appears not as just an emblem of Realist philosophy or of the human yearning for that which is absent, but as an objective correlative of the intersubjective experience, in both its succesful and failed versions. This reading is dependent on recognizing the importance of triangulated atention in the novel. 162 Chapter 6: The Editorial Conversation Criticism in Practice I move now from interpretive criticism to a discussion of textual criticism. Where the last chapter provided evidence that a social-cognitive framework can open up new readings of individual texts, this chapter wil addres the benefits that it can provide for curatorial considerations regarding a given text. Textual criticism concerns itself with the project of producing scholarly editions of texts. In formulating guidelines for how this should be done, what properly constitutes the task of the scholarly editor or curator, and what the desired form of the editor?s product should be, it therefore grapples with underlying questions of what constitutes a text and its meaning. I mean to criticize the justifying naratives that underlie both of the prevailing approaches to textual criticism, the ?final authorial intention? approach to textual criticism, asociated with W. . Greg, Fredson Bowers, and G. Thomas Tansele, and the ?social text? approach, most famously asociated with Jerome McGann. This critique is a return to the theoretical questions addresed in chapter two of this study, applied to the concrete concerns of editorial practice. When people, whether members of the profesion or of the laity, read, they have a working theory of what they are doing. There sem to be two primary theories available to them at the moment. The first is fundamentaly romantic, and holds that there is a ?real?, fixed meaning that 163 they can rely upon and which can be located in the privileged intentions of an author or the privileged interpretations of an ideal reader. The other is fundamentaly behaviorist, and holds that only the realized, physical, perceptible traces of a text are fair game; any supposition about the conceptions that lie behind it is as taboo as an appeal to introspection would be to behaviorist psychology. Both of these approaches are broken, because there is a real problem with the conception of meaning these theorists have to work with. Neither is faintly realistic if one believes in the reality of concepts and meanings; but they also run aground as long as one believes that readers frequently believe in these things. The ?eclectic text? or Greg-Bowers-Tansele school of textual criticism, takes the fundamentaly romantic view of the nature and meaning of a text, postulating the existence of an abstract, ideal ?work? which is but brokenly reflected in the mere earthly expresions of that work that are realized in the world. The social-text approach, by contrast, takes a studiously behaviorist approach to the textual encounter, warning the prospective editor that it is impossible to curate a text without also interpreting it, but urging as much as possible that she pursue an understanding of texts as ?a laced network of linguistic and bibliographical codes? (McGann 1991:13) which should be compiled as completely as possible to provide a comprehensive record of the history of a text. The social text approach is philosophicaly appealing, in that it captures the complicated nature of authority and textual construction. It suffers, however, from its atempt to legislate a rationalist, anti-intentionalist approach to editing and reading. While there are excelent reasons to resist appeals to authorial intention for 164 constructing and reconstructing the meaning of a text, the reality is that many of the people involved in the socialy situated, socialy constructed production and contextualization of a text wil themselves be relying on intentionalist strategies of interpretation. Further, practical exigencies mean that authorial intention and the work/expresion framework wil guide the decisions of editors and archivists for the foreseable future. A cognitively realistic approach serves to bridge the two approaches, and the mechanics of joint construal provide a productive framework for talking about these proceses without obscuring either aspect of the facts on the ground. This chapter ilustrates some shortfalings of these two approaches to the curatorial project through a social-cognitive analysis of the stylistics, textual history, and editorial reception of one of the most famously revised poems in modern English, Marianne Moore?s ?Poetry.? While the capacity for revision is often considered a characteristic quality of writen discourse, face-to-face conversation frequently involves revision and repair. I argue that the thematization of omision in ?Poetry? is best analyzed as a joint construal of both the poem and its history, built up together by author, editors, publishers, critics, and other readers. This approach provides an opportunity to look at the longitudinal stylistics of a literary work as a discourse unfolding over time, and iluminates the connections betwen these literary acts and their counterparts in natural discourse. 165 Editing and Conversation Writen language?s capacity for revision, or ?working over,? is often presented as one of the primary characteristics that distinguish it from spoken language (se, e.g., Chafe 1994) 12 . But face-to-face conversation, despite its evanescence, also involves proceses of revision, often enacted collaboratively betwen multiple discourse participants. Indeed, recent work in cognitive science suggests that language, human beings, or both are optimized for constructing meaning interactively and iteratively. What constitutes ?authority? in conversation often turns out to be as complicated as the models of textual authority used by critical theorists and literary historians. While the comparison of revision in talk and revision in print has interesting implications for the study of both literature and conversation, in this chapter I wil concentrate mostly on the ways that this comparison can inform and enrich our reading of literary texts. Specificaly, I wil provide a detailed analysis of relevant aspects of social cognition as they play out over the textual history of one of the most 12 The quality described in these works is a feature of the relatively non-ephemeral nature of writen texts, not a unique consequence of their being writen. Paterns of embedded repetition and paralelism (?ring composition?) characterize many oraly transmited works, providing a fixed structure despite the lack of a physicaly fixed medium. These configurations can be extremely elaborate and strict in their arangement. The Zoroastrian Gathas, for example, composed around 1200 B.C.E. and transmited oraly for hundreds of years, include pervasive and highly complex paterns of chiasmus and other concentric symmetries, described in Schwartz 1986 and Douglas 2007. The survival of this sort of poetic structure reminds us that the spoken modality can provide for enduring, and hence in this sense revisable, texts. Similarly, audio and video recordings can be ?read,? reviewed, and edited??worked over??by those who produce and diseminate them. 166 famously revised poems in modern English, Marianne Moore?s ?Poetry?, and make a case for reading this history as a complex of multiple joint activities. This approach provides an opportunity to look at the longitudinal stylistics of a literary work as a discourse unfolding over time. It also provides a new ay of thinking about Moore?s ?thematization of omision? (Peterson 1990). This has been alternately treated as a loss or mutilation, as in Kenner (1967), an expresion of ambivalence and alienation, as in Gregory (1996), or the source of a fundamental conflict betwen the author?s ?editorial wishes and the driving need to represent her work? (Schulman 2003:xxi). Close examination of the publication and editorial history of the poem reveals that these revisions were neither deletions nor ambivalent; they represent multiple turns of an extended discourse among multiple participants. It is as a discourse-level phenomenon?that is, as something that transpires over several turns of this editorial ?conversation??that certain features of the poem emerge, especialy much of its wit. Further, this approach iluminates the role that reading editions, perhaps even especialy the kind of reading editions that ?social text? approaches to textual criticism typicaly abhor, play in creating the actual social text of a literary work. Revision and ?Poetry? While Marianne Moore?s productive years spanned the majority of the twentieth century, the poems first published betwen 1920 and 1935 are widely agred to constitute her most succesful and influential work. These fiften years 167 embrace two distinct phases in Moore?s oeuvre, divided by a gap of about six years in which her atention was largely devoted to editing the prominent literary and cultural magazine The Dial. Many of Moore?s contemporaries were vocal in their praise of her work, including Ezra Pound, Wiliam Carlos Wiliams, and T. S. Eliot, as wel as her sometime editors H. D. and Walace Stevens. Moore?s poems are known for their precise, concrete imagery, surprising juxtapositions, and sylabic stanza paterns, in which syllable count, rather than the number and arangement of streses, determines the length of lines. During the early 1920s she experimented briefly with fre verse, but soon returned to the syllabic stanzas of her earlier carer. The metrical regularity of these stanzas is not readily apparent to the ear, as it is counterbalanced by the iregular arangement of streses. Their paterns are, however, imediately evident on the page, where the idiosyncratic but repeating arangement of lines of varying lengths is both made visible and reinforced by the deployment of indentation. These syllabic arangements were arived at by a proces in which the overal stanza form for any poem grew out of paterns that had emerged in an earlier, unplanned stage of writing (Holley 1984). A pleasing arangement of lines would then serve as a template for other stanzas: ?I never ?plan? a stanza. Words cluster like chromosomes, determining the procedure. I may influence an arangement or thin it, then try to have succesive stanzas identical with the first? (Moore 1969:34). Thus Moore?s stanza schemes vary greatly from poem to poem, and across diferent versions of the same poem. The 1934 version of ?The Frigate Pelican,? for example, has five-line stanzas of the syllable patern 15-12-1-9-9. The 1967 version changes 168 to four-line stanzas with much shorter lines, following a 7-9-7-6 syllable patern. A similar range of variation obtains over her entire oeuvre. Moore maintained that her unusual methods of stanza construction were intended at least in part to create ?natural? or ?conversational? efects (Schulman 1986; Moore and Schulman 1969). Despite Moore?s claims, however, many critics have sen her formal techniques as part of a deliberate efort towards obscurity. Gregory (1996:144), for example, writes that Moore?s poetry features ?a slew of strange and estranging devices that she employs on purpose to make reading dificult.? These potentialy alienating devices include not only her use of syllabic verse, but also ?her constant revision of poems over versions (which makes it dificult both to determine what constitutes the poem and to posses it in memory).? While it is far from clear that her purpose in doing so was always or even usualy ?to make reading dificult,? Moore was inded prolific in her revisions. The most extensive and striking of these were those applied to the poem ?Poetry?. It went through at least six major iterations and many dozens of printings over sixty-thre years, shrinking from five stanzas to a single stanza of thirten unrhymed lines, growing again to thre stanzas and fiften lines, and finaly truncated to a mere thre lines, with a variation on the original five-stanza version included in an endnote. Variations on the five-stanza version were published many more times in Moore?s lifetime than any other version, and she went on revising this longer version over this entire forty-eight year period, making various smal adjustments right up until the final endnote version (1967:266-7), which reads: POETRY (page 36) 169 Longer version: I, too, dislike it: there are things that are important beyond al this fiddle. Reading it, however, with a perfect contempt for it, one discovers in it, after al, a place for the genuine. Hands that can grasp, eyes that can dilate, hair that can rise if it must, these things are important not because a high-sounding interpretation can be put upon them but because they are useful. When they become so derivative as to become uninteligible, the same thing may be said for al of us, that we do not admire what we cannot understand: the bat holding on upside down or in quest of something to eat, elephants pushing, a wild horse taking a rol, a tireles wolf under a tre, the imovable critic twitching his skin like a horse that fels a flea, the base- bal fan, the statistician? nor is it valid to discriminate against ?busines documents and school-books?; al these phenomena are important. One must make a distinction however: when dragged into prominence by half poets, the result is not poetry, nor til the poets among us can be ?literalists of the imagination??above insolence and triviality and can present for inspection, ?imaginary gardens with real toads in them,? shal we have it. In the meantime, if you demand on the one hand, the raw material of poetry in al its rawnes and that which is on the other hand genuine, you are interested in poetry. This five-stanza poem stands in stark contrast to the version that appears in the main body of the collection (1967: 36): POETRY I, too, dislike it. Reading it, however, with a perfect contempt for it, one discovers in it, after al, a place for the genuine. 170 Even more versions of the poem have appeared since Moore?s death. These include scholarly editions that present partial or complete ?variorum? overviews of the entire history of the poem, as wel as numerous editions in anthologies designed for students and casual readers. The later most comonly present a five-stanza version in the main text, sometimes acompanied by a footnote or endnote that includes the thre-line version. The long period over which these many printings and revisions appeared is often treated as a period in which Moore?s degre of public renown and influence with other poets increased, but her own performance as a poet was in decline (cf. Molesworth 1990; Wilis 1990). However, this characterization of her later output relies on problematic asumptions about what kinds of activities constitute ?writing? or ?doing? poetry. Even if one agres that the poems first published in the 1940s, 1950s, and 1960s were leser work, this judgment does not acount for Moore?s continual refinement of her earlier poetry. The revision and printing history of ?Poetry? and several other wel regarded poems that made their first appearance before 1935 are such that these poems were unfixed, changing, stil-in-construction works until at least 1967, thirty-odd years after the first versions appeared in print. Omisions Are Not Acidents (But They Aren?t the Whole Story, Either) The quantity and magnitude of Moore?s revisions present a special chalenge to scholars of her work. There are many authors for whom revisions are largely a 171 mater of editorial interest: a single definitive edition with a few explanatory footnotes is perfectly adequate for scholars not specificaly concerned with publication history. But no one who hopes to write seriously about Moore?s poems can ignore her revisions and re-revisions. They are simply too numerous and, in many cases, too dramatic to be glossed over. Furthermore, many of her poems take poetry and communication themselves as their subject, making the proces of their construction particularly relevant to analysis of their structure and content. Litle surprise, then, that critics agre that Moore?s poetics are inextricably ?entwined with their printing history? (Kenner 1969: 161). Extremely detailed acounts of the complete publication history of these variants can be found in Honigsblum (1990) and Schulze (2002). Such a complete transcript would take up far more room than I have here; instead I provide a general overview in Table 5-1 (which is also indebted to Peterson (1990:237)), and wil take a closer look at variations and paterns of variations as they arise. This chart covers al major appearances outside of anthologies and the first appearance of each version, with the total number of printings of each version in Moore?s lifetime, acording to Honigsblum (1990), noted alongside its first appearance. This summary necesarily occludes many smal variations. For example, there are many minor diferences of punctuation across the fifty-five printings cited at variation D, below. However, it does give a good sense of two important facts about the poem?s history: First, even in Moore?s lifetime, variations on the five-stanza version were by far the most frequently printed. Second, while at first glance the overal trajectory of these revisions sems to be one of abbreviation, which is also 172 how the history is generaly represented in critical responses to the work, the reality is significantly more complicated. Publication Date Stanzas: Lines Printings A. Others, 5, No. 6. July 1919 5:30 3 B. Others for 1919 1920 5:30 C. Poems (reverts to 1919) 1921 5:30 D. Observatons Dec 1924 5:29 (plus note) 55 E. Observatons March 1925 1:13 1 F. The New Poetry 1932 3:15 5 G. Selected Poems (Macmilan) (reverts to 1924) 1935 5:29 H. Selected Poems (Faber) (reverts to 1924) 1935 5:29 I. Collected Poems 1951 5:29 J. Collected Poems 1967 1:3 / 5:29 10 K. Complete Poems 1981 1:3 / 5:29 L. Complete Poems 1994 1 M. Becoming Marianne Moore 2002 Variorum N. The Poems of Marianne Moore 2003 5:29/variorum Table 5-1: A sumarized edition history of ?Poetry? The publication history of the poem is distinctly marked not only by deletions but also by substitutions, additions, and reversions. Often neglected in both histories and critical editions of the work, for example, is the period betwen 1925 and 1932, in which Moore first abandoned and then reinstated a stanzaic structure. The 1925 edition is a fre-verse version of just thirten lines: POETRY I, too, dislike it: there are things that are important beyond al this fiddle. The bat, upside down; the elephant pushing, a tireles wolf under a tre, the base-bal fan, the statistician? 173 ?busines documents and schoolbooks?? these phenomena are pleasing, but when they have been fashioned into that which is unknowable, we are not entertained. It may be said of al of us that we do not admire what we cannot understand; enigmas are not poetry. It is unclear what led Moore to make this experiment in fre verse, especialy considering how vigorously she had defended her preference for syllabic stanzas over fre verse in leters and her conversation notebook just a few years previously (qtd. in Honigsblum 1990:187), complaining of an acquaintance?s ?narow minded? preference for fre verse, but demurring, ?wel?I?ve never writen fre verse and don?t know how to write it.? Perhaps this revision was an atempt to repair that state of afairs. If so, she did not sem to be long satisfied with the result. During the period before she became editor of the Dial, around the same time that she was producing the thirten-line ?Poetry,? she made similar revisions to other poems originaly drafted in syllabic stanzas, for example ?A Grave? and ?When I Buy Pictures?. Afterwards, she turned away from fre verse for the duration, and a mark of this rejection was the publication of a yet new version of ?Poetry? (1932), this time with syllabic stanza structure both restored to and altered from the original: POETRY I, too, dislike it; there are things that are important beyond al this fiddle. Reading it, however, with a perfect contempt for it, one discovers that there is in it, after al, a place for the genuine: hands that can grasp, eyes that can dilate, hair that can rise if it must, 174 the bat holding on upside down, an elephant pushing, a tireles wolf under a tre, the imovable critic twitching his skin like a horse that fels a fly, the base-bal fan, the statistician?nor is it valid to discriminate against busines documents, school-books, trade reports?these phenomena are important; but dragged into conscious oddity by half poets, the result is not poetry, This we know. In a liking for the raw material in al its rawnes, And for that which is genuine, there is liking for poetry. Here we se the return not only of syllabic stanzas but also of the place for the genuine, the hands that can grasp, eyes that can dilate, the raw material in al its rawnes. Gone are the quotation marks around ?busines documents? and ?school- books?; for the time being they are simply present, not spoken. Meanwhile, the stanzas of the original version have returned, but their form and line breaks are altered: from six lines they have been truncated to five; the changes to the syllabic- line patern diverge from both extremes of the original. No line in this version is as long as the longest in the five-stanza version, and none as short as the shortest, so that the punctuation of the line breaks is les syncopated and more regular than that produced by the juxtapositions in the prior approach. Thus this revision features al thre kinds of changes in plenty: some deletions, but also many additions and reconfigurations; over this period Moore is tweaking, altering, and restoring. This fiften-line, thre-stanza version apparently satisfied Moore and her anthologists off and on through 1952; it appeared with her blesing in Hariet Monroe?s anthology The New Poetry in 1932, 1934, and 1935 editions, and Louis Zukofsky?s A Test of Poetry in 1948 and 1952. But toward the end of this same 175 period, in her 1951 Collected Poems, Moore returned to an even longer version, nearly identical to the 1924 edition. Thus the history of this poem and Moore?s revisions to it comprise a trajectory that is not particularly acurately described as a diminution. Indeed, as Table 1 shows, the full history of the poem includes several cases of ?restorations? of longer versions after the presentation of shorter versions. What?s more, the shortest version of the poem never appears entirely alone. Nor do alterations of smaler kinds follow a simple path from verbosity to concisenes. The comma before ?too? in line 1 of al versions, for example, comes and goes. Yet it is undeniable that this narative of progresive atenuation has taken hold in the critical consciousnes, in part because of Moore?s explicit encouragement. Following Moore?s example, acounts of this history primarily focus on acts of diminution and exclusion. Critics have characterized Moore?s edits variously as ?revision by subtraction? (Wilis 1986), ?revisionary chopping? (Hicok 2000:493n37), or a series of ?drastic fits of rectitude? (Kenner 1975:107). Moore semed to take special pleasure in teling friends about the ?drastic cut? (Gregor 1984) she had made for the 1967 volume, and savoring their shocked reactions. This atitude was reinforced concisely in her epigram to Complete Poems, ?Omisions are not acidents.? This line is expertly crafted to enforce a particular construal of the versions found in this edition; not only does it lay emphasis on those revisions that remove rather than add, it also characterizes those alterations (even, as we wil se, some that might more acurately be described as additions) as omisions, as lacks, as empty spaces where something else ?ought? to have been, encouraging readers to search for favorite poems and sections that do not appear. 176 An emphasis on acts of omision, rather than of commision, also harmonizes more satisfyingly with other recurring themes in Moore?s poetry. She often took as her subject the smal, the disprefered, and the absent. The speaker of her 1924 poem ?Silence,? for example, reports that her father ?used to say? that ?The deepest feling always shows itself in silence; / not in silence, but restraint.? The poem itself sems to take these words to heart; al but two and a half of the fourten lines devoted not to the speaker?s own voice, but to quotation. The speaker offers no direction on how to take the father?s lecture, restraining her commentary to only the thre lines ?My father used to say,? ?Nor was he insincere in saying,? and ?Inns are not residences?. Similarly, ?Poetry? itself presents a mystery of omision in its first line: ?I, too, dislike it.? With whom is the speaker alying herself? The reader? The many people who?disliking poetry as they do?wil never be her readers? ?Sensible people,? as Kenner (1975:106) suggested? Critical responses to ?Poetry? thus often center on diferent approaches to this ?thematization of omision? (Peterson 1981). There are thre basic takes on the phenomenon. One is to treat the sequence of revisions as damages or losses, another is to treat them as additions or afirmative statements, and the third is to treat them as an expresion of, or enactment of, ambivalence. Hugh Kenner is a particularly dramatic perpetrator of the first approach, refering for example to the five-stanza version as ?the one scared by al those revisions? (1967: 1432). In this conceptualization, many variations published under the title ?Poetry? are compresed (Fauconnier and Turner 2002) into a single, concrete entity that the poet has altered many times. This entity is also metaphoricaly characterized as a living body, and the 177 alterations that remove material from that body as violent mutilations. In this way, even a new, intact printing of an earlier version can be ?scared? by the publication of shorter variations. The notion that Moore?s revisions somehow damage or detract from the five- stanza version follows naturaly from a highly pervasive way of thinking and talking about texts; one which, indeed, I have not been able to avoid completely myself in the discussion above. It is very common to compres analogies betwen many specific individuals into an identity relation, producing a unique individual in the resulting blend. Disanalogies across those specific individuals are then understood in terms of changes to that unique individual. Everyday language is full of expresions for compresing disanalogy into change in this way. ?Every year my cel phone gets smaler and my bil gets bigger,? for example, presents a compresed blend in which there just two items, changing their size over time. Understanding the blend involves unpacking this compresion, recognizing that there are several distinct phones and equaly distinct phone bils. Kenner?s characterization is thus understandable, but problematic. After al, nothing has actualy been destroyed?many copies of the original printings of previous versions exist, and there is nothing stopping anyone from re-reading or even re-printing whatever version they please, and the very collection that presents the traumatizing thre-line revision includes a complete instantiation of the ?scared? five-stanza version in every copy. What?s more, the variation that is surely most in danger of vanishing from readers? experience is not the five-stanza version, but the thirten-line version in fre verse, which was apparently a one-time experiment 178 appearing only in the second (1925) edition of Observations. The thre-stanza version has not fared much beter, published only five times in Moore?s life, in thre editions of an anthology edited by Hariet Monroe and Alice Corbin Henderson, The New Poetry (1932, 1934, 1938) and two editions of Louis Zukofsky?s anthology, A Test of Poetry (1948, 1952). Stil, even these versions, which had al but vanished in the last decades of the twentieth century, do remain in print to some degre, quoted in critical esays and in the end notes of Grace Schulman?s epic collection The Poems of Marianne Moore (2003), as wel as Robin Schulze?s (2002) variorum edition of Moore?s early poetry. A contrary line has been to take the history of revisions as a kind of acumulation. M. L. Rosenthal (1969:127) writes of a progresion of afirmative statements that invite readers to appreciate the relationship betwen an extended poetic work and the sum of its parts. He suggests that the 1967 volume is ?playing with the two versions? concurrently, ?dramatiz[ing] the insistent fact that many poems are comprised of what we might cal a kernel and a context, and that often, in fact, the kernel does reside in the first line or lines.? Others take the sequence (again, generaly concentrating on only the longest and the shortest variations) as an expresion of, or enactment of, ambivalence. Bonnie Costelo, for example, argues that ?[t]he two versions stand not as original and revision but as two alternative statements.. Revision, whether within the text or betwen texts, is an esential part of Moore?s aesthetic. It is motivated by an esential ambivalence about poetry?s capacity to asert? (1981:25-26). This last view, I think, is more right than the position that new versions detract from or alter old ones; they may supercede earlier 179 versions, entirely or in part; or they may coment on them, or co-exist. But there are problems with al of these characterizations. First, it is a vast simplification to talk about something that was enacted over the course of many years as either a unitary object or a simple set of tokens. The revisions are not independent of one another, nor were they constructed in isolation in the interpersonal sense. Instead, they are parts of an extended, multi-turn interaction. Further, it is only by taking a very selective view of the entire history of these revisions that this history appears to be solely or primarily a mater of omision, or even a collection of increasingly pared-down iterations. That view requires one to bracket the history very selectively, taking only Moore-initiated versions, and not even al of those, into acount. The fate of the thirten-line version, for example, represented a temporary paring down, followed by a return to the longer form of the five-stanza version. The later persists in its relative maximalism across the entire publication history of the piece, and is reproduced in anthologies and other reading versions to this day. And yet it certainly is true that omisions are increasingly highlighted in later versions of the work published in Moore?s lifetime, as wel as in posthumous versions. Moore?s 1967 epigram of course profiles acts of omision, and she also often deployed notes and other paratextual materials that cal atention to cuts in the text and downplay other kinds of revisions. Critical responses and editorial annotations follow suit. In the end, then, omisions are thematized, it sems, even if this thematization does not exactly reflect the ?true? history of the poem. How best to acount for this state of afairs? 180 Joint Construal and Revision in Conversation It is tempting to say that the omision narative that Moore, her editors, and her critics endorse is simply inacurate. It selectively ignores real events in the manuscript and publishing history of the poem, and it suggests that both Moore?s intentions and the material history of the poem have always been coherently ?about? omision, even though a look at the full publication history makes it clear that the actual trajectory was much more vacilatory and complicated. Al of this is true. But it would be a mistake, I think, to argue that critics should abandon this story of the poem. Instead, we should recognize the thematization of omision as a joint construal of ?Poetry? and its history which the author, editors, publishers, critics, and other readers have built up together?a construal that Moore initiated, but which was cemented by al of these participants, including editors who constructed their own reading versions of the poem after her death. The term joint construal comes from Clark (1996), and refers to the way that participants in a conversation try to establish a shared understanding of what the speaker of each utterance in the conversation ?is to be taken to mean? by it (p. 215). Sometimes this happens smoothly and without incident, but it always requires contributions on the part of both speaker and hearer. In Clark?s terminology, the speaker proposes a joint project and the hearer takes it up. The uptake response provides evidence that the hearer understood the speaker?s utterance, and also tels the speaker something about how the hearer is construing that utterance. For example, if Ema says to Frank, ?Have some pie,? each of the responses in (22) would suggest a diferent construal of that utterance: 181 (22) a. Yes, ma?am! [an order] b. No thanks, I?m fine. [an offer] c. What a good idea! [an advisory] If the displayed construal matches Ema?s original intentions, she can proced acordingly. Her next turn, then, ought to reflect her sense that her proposed project has been succesfully taken up, giving Frank cause to recognize that this construal is now jointly held betwen them, and so forth. If the construal doesn?t match, she can correct it (?I?m not asking you, I?m teling you!?) or leave Frank?s construal unchanged and revise her own intentions about what she should be taken as doing. Sometimes it happens that a number of conversational turns are required to establish a jointly acepted meaning. Both the original speaker and the hearer may contribute to the meaning being constructed, as in this example from the Michigan Corpus of Spoken Academic English, speaker labels altered for ease of reference: (23) 1D: so for thirty-thre eighty-thre, if I use these primers, and I use the twelve sixty-thre A probe 2C: you get right regulation [continuation of 1 by hearer, C] 3D: right regulation [acceptance of 2 via repetition] 4C: yeah [acceptance of 3] (MICASE MTG400MX008) In other cases, the joint construal that emerges by the end of a sequence is significantly diferent from the speaker?s original intended or expresed meaning, or even from both the speaker?s and the hearer?s original construal of the joint conversation project. In (24), for example, thre interlocutors work their way to an 182 entirely new shared construal over multiple turns of revision, incorporating contributions from al participants: (24) S1: aces a culture, is, i don't quite know what that means. S2: replicate would that be beter? S1: replicate is, replicate a culture, no. beter expresion, Ken? S3: um, they might emulate it. i mean they're looking at the ge- the geography of te- of Ptolemy and Pliny the Elder at this point and going_ they they ex- they explained it in the past, they were right because they ruled an entire empire. S1: so your suggestion is that, is that, we might say, that they, um.. they looked, to the past, and they emulated, the example, of the past. now, let?s just

emulate?s a prety good word (MICASE DIS315JU101) In these examples, neither the structure nor the meaning of an utterance is fixed by a speaker?s original expresed intentions. Instead, construals are jointly negotiated over the course of the conversation. Joint construals of this sort are not merely the product of smal, local slices of discourse such as those presented in these atested examples. Indeed, they multiply over more protracted discourse and larger discourse communities, as in the case of ?Poetry?. The thematization of omision in ?Poetry? is not something that can be located in any single version of the poem. Neither does it arise out of the series of published revisions of the poem; it is not simply a fact or falsehood about the revisions that Moore enacted and published. Instead, the thematization is a construal of these ?uterances? of the poem, one that has been worked out largely outside the textual history of Moore?s revisions. There are thre major avenues by which this construal has been solidified. First, a number of individual extra-textual conversations in which Moore played a major role have been particularly influential by virtue of their frequent and 183 prominent repetition. A striking example is a story that Moore liked to tel about herself and her longtime editor at Viking Pres, Edwin Kennebeck. This exchange is quoted, paraphrased, and reported on many times in print (e.g. Schulman 1969; Gregor 1984; Peterson 1990; Schulman 2003) and appears to have made the rounds in person, as wel, so that it eventualy became a clasic story that one told about Mis Moore and her iconoclastic ways. Here is one version of it: Grace Schulman: Have you changed many of the poems for this edition? Marianne Moore: Yes, I have changed them somewhat. Edwin Kennebeck, r. Kennebeck, said, ?Marshal Best [a senior executive at Viking Pres] is going to fal dead when he ses ?Poetry? reduced to thre lines??. Mr. Kennebeck said, ?Oh no. But I think some people are going to complain if you leave the whole thing out.? But then he said, ?Wel, I thought of this: How would it be if we had an appendix and put that in the back, together with the other things you have reduced to nothing?? ?Wel,? I said, ?that?s fine. Then it saves the serious reader from looking up these things as they were.? (Schulman 1969: 160-61) Note that this reported exchange presents a very particular understanding of what Moore ?is to be taken to mean? by the 1967 revision, along with the suggestion that several diferent people concur in this construal. Marshal Best is expected to react dramaticaly when he ses that the poem has been reduced to thre lines. Kennebeck proposes an appendix containing the five-stanza version, which is described as one of many ?things? that have been reduced to nothing. Moore endorses this construal not only by repeating it for Schulman?s ears but also in her own reported reply, in which she refers to the long version as an example of things as they were. This one pasage 184 economicaly puts forward both a concise display of the omision narative, complete with elision of many of the other stages in the revision history of the poem, and the suggestion that this understanding not only is held by the speaker, but is part of the common ground for many, perhaps everyone, in the smal, informed group asociated with the publication of ?Poetry.? The frequent reteling of this story connects to the most obvious means by which the omision narative is performed and ratified: simply, by repeated displays of a certain understanding of the poem and its history. These are exhibited in critical esays like those surveyed in section thre of this esay, as wel as in editorial notes, biographies, and the like. The reiteration of a given construal is a common feature of the joint enactment of new or revised construals in conversation. In addition to the kind of word-for-word repetition that serves as a signal of confirmation in examples (23) and (24) above, this proces is most reminiscent of conversational uptake responses. Uterances such as ?Omisions are not acidents? and the Kennebeck story constitute a proposed joint project as much as any turn at conversation. When editors and critics respond, they are choosing whether or not to take up that proposal, to provide a counter-proposal, or to find some middle ground betwen the two. Among the most interesting sites of interpretive construal are posthumously constructed reading versions of the poem, such as those that appear in the Norton anthologies of poetry and American literature, the Oxford Anthology of American Literature, and Schulman 2003. These typicaly present the five-stanza version in the main text of the volume, and the 1967 thre-line version in notes, often in a (marked) footnote rather than an (unmarked) endnote. This approach creates an omision 185 narative in miniature, in which the five-stanza version is repeatedly construed as the originary and ?true? version, and the thre-line revision is presented as a surprise second act. The relative status of these two versions is additionaly reinforced by the fact that the most frequently quoted lines from the poem, such as ?real gardens with imaginary toads in them,? can al be found in the five-stanza version, and are strikingly absent when the reader turns her atention to the footnote or turns the page to the endnote. Thus, editions for a more popular audience, which neither provide a full textual history nor provide one of the Moore-endorsed presentations of the poem?nor even presentations matching any created in Moore?s lifetime?participate significantly in the textual conversation, contribute to a public joint construal of the work, and conspire to reinforce and re-inscribe the thematization of omision. Eclectic Texts and Social Texts The perhaps surprisingly substantive role that idiosyncratic reading editions play in enacting the socialy constructed meaning and composition of this poem connects to a larger debate in Anglo/American editorial scholarship over two influential approaches to the project of creating scholarly editions and how to think about ?authority? in a text. The first of these is the ?final authorial intention? tradition asociated with W. W. Greg and Fredson Bowers, in which the job of the scholarly editor is to create a single text that approximates as closely as possible ?the form of [the author?s] work he wished the public to have? (Tansele 1976: 167). That is, the object of the editor is 186 to come as closely as possible to producing a definitive text reflecting an author?s final intentions. G. Thomas Tansele?s work grounds the Greg-Bowers methodology in a theory of texts that distinguishes betwen a ?work? and ?expresions? of it. The ?work? is an abstract entity, the form of a piece of literary art that exists only as an ideal. Any realized instantiation of an artwork?in print, performance, manuscript, or otherwise?is merely an ?expresion? of that ideal. By its very nature, then, every reproduction is a mere approximation of the true, ideal work. In this school of textual criticism, the job of the editor is to approximate the ideal as closely as possible. More recently, editorial theory within the academy has largely come to favor a ?social text? approach, asociated most famously with Jerome McGann (1983, 1991), in which authority rests not with a solitary author but in a social proces of textual production involving editors, typeseters, proofreaders, censors, anthologists, and others. This approach difers from the former not only in its fundamental conception of the nature of a ?text,? but also in the final products that it aims to produce. The goal of editors working in the Greg-Bowers-Tansele school of textual criticism is to construct a single, stable, and readable, authoritative edition. (McGann criticizes these editions as ?eclectic texts? that have no historical counterpart.) Editors who favor the social text approach advocate, instead, a ?variorum,? an acount of an unstable textual history, with atention to the social and material contexts of al the variants of a given work. While the social text tradition rejects both the theory and methodology asociated with Tansele?s work/expresion distinction, that distinction is alive and wel in the practice of both editors and librarians, as it is enormously useful for 187 practical purposes of cataloging, publishing, and teaching texts. The International Federation of Library Asociations and Institutions, for example, makes use of these categories in its Functional Requirements for Bibliographic Records (1998:16-18, italics in original): ?A work is an abstract entity? Relating expresions of a work indirectly by relating each expresion to the work that it realizes is often the most eficient means of grouping related expresions.? Similarly, publishers such as W. W. Norton in the United States do a booming busines in commisioning and producing demi-scholarly editions of clasic texts for the undergraduate market. These teaching editions typicaly hew very closely to the ?final authorial intention? approach, for the simple reason that it is a tried and true way of packaging a text to make it amenable to reading. My joint construal acount is in substantial sympathy with the notion that literature is composed not of abstract ?works? and their imperfect material ?expresions,? but of a mesy aray of socialy enacted textual productions. However, this analysis ascribes a much higher importance than the social text approach would grant to the kind of reading editions produced by editors and publishers working either explicitly or implicitly in the work/expresion tradition. The diferences become particularly evident on examination of McGann?s 1992 discussion of ?Poetry,? which he takes to be an excelent ilustration of the failings of the ?final authorial intention? approach to texual criticism. McGann points out, correctly, that while the 1967 thre-line version of the poem evidently represents Moore?s ?final intentions? towards the towards the piece, ?it is equaly clear that the earlier and longer work wil never be superseded by the 188 later revision; indeed, the peculiar force of the revised version depends in important ways upon our knowledge and recollection of the earlier? (1992:86). So far, so good. However, he goes on to argue that editors and readers would ?probably do wel to regard it as a new and separate poem rather than as a revision of the earlier work.? I disagre whole-heartedly with this characterization. The social text tradition would be beter served by an understanding of the role that both their own analysis and that of the work/expresion school play in forming a conversation around textual works; in the case of ?Poetry,? the prominence of reading editions makes them perhaps the most important participants in the creation of a joint construal of the poem after the revisions and comments of Moore herself. The formation of a social text does not end at the point where those engaged in the proces of textual criticism enter the conversation. Thematizing Trouble, Thematizing Repair Finaly, I would suggest that a fuler appreciation of the dynamics of revision and joint construal that have characterized the publication, disemination, and reception of ?Poetry? clarifies more than the poem?s bibliographic history. It can also open up our reading of the poem itself. As it happens, there are features that persist across al versions of the poem that play not only with the notion of omision, but specificaly with linguistic resources for revision in conversation, creating a special sense of both frustration and conspiracy betwen the reader and the speaking persona, 189 and highlighting the ways that poets and poetry both engage with and decline to engage with their readers. In conversation analysis, the term of art repair is used to describe what happens when speakers identify something in a conversation as a source of ?trouble.? In the seminal paper on repair in conversation, Schegloff, Jeferson and Sacks (1977) note that sources of trouble are not limited to erors, and responses to trouble can take forms other than correction. For example, dificulties in word recovery often trigger repair sequences. One speaker may supply a word that is eluding the other, or a speaker might first search for a word and then supply it herself. In other cases, no noticeable eror occurs at al, as in examples (23) and (24) above. In sum, the phenomenon in question is les one of ?repair? per se than it is one of revision; conversation alows speakers to add to, subtract from, or make substitutions for elements that have already been introduced into the stream of talk. Acts of revision in conversation can be usefully divided into two minimal parts, initiation and outcome. The participant who produces, or enacts, a revision is not necesarily the same as the one who initiates it. Repair initiation cals atention to, or ?undertakes to locate? (Schegloff 2007:101) a ?trouble-source? in a piece of talk, while repair itself undertakes (idealy) to eliminate the trouble. Whatever version of ?Poetry? we choose to start our reading, the very first line presents a striking trouble-source, namely, ?I, too, dislike it.? Saul Kripke (1990) has observed that too is an example of a peculiar subset of presupposition triggers that are more demanding than most. Ordinary presupposing constructions can be used informatively; that is, they can cary presuppositions that are not already part of the 190 common ground and be felicitous nonetheles. For example, if I say, ?I have to pick up a package for my husband? to someone who had no idea whether or not I had a husband, my interlocutor can easily acommodate (Lewis 1979) the presupposition triggered by my husband (namely, that there exists a person who fils this role), adding it to the common ground as it is introduced into the discourse. Kripke points out that too does not sem to permit this kind of acommodation. The presupposition triggered by too does not apply only to the common ground but to facts about the conversational history itself. Consider the following example: (25) John can?t come tonight. He?s having dinner in New York, too. The expresion too in combination with the appropriate prosodic structure in the second sentence of (25) triggers not only the presupposition that someone other than John is having dinner in New York tonight (which is trivialy true) but also that someone salient to both interlocutors had dinner in New York last night. Furthermore, and most importantly, it triggers the presupposition that this fact was either mentioned explicitly in the previous discourse or is easily recoverable from it. This last aspect of the presupposition is what makes it ?anaphoric.? The literature on anaphoric presuppositions in philosophy of language and pragmatics naturaly presents examples like (24) in terms of what they, or their hearers, cannot acommodate. This does not mean, however, that speakers never produce an anaphoric presuppositional trigger like too without first producing the prior utterance that should be required license it. Instead, when they do produce them 191 in the absence of their licensing context, they or their interlocutors are likely to initiate a repair sequence: (26) S4: Zack was [asking for grapefruit] too. S6: what? S4: you said you wanted some. S6: no, i don't want any. i said it was a bad addiction that he was geting into. S4: oh (MICASE LES320SU085) Similarly, the absence of an antecedent to the first line of ?Poetry? serves to provoke the reader in ways that go beyond the more conventional deployment of, for example, definite reference or pro-form in the first line of a literary work. It is very common for Western literature to do the later, with the efect of ?conveying that [a given piece of information] should be mutual knowledge? (Hobbs 1987:7). The more exotic use of too in ?Poetry? suggests that there must be utterances prior to this one, and in their absence these create a trouble source that cries out for repair. Thus the highly abbreviated thre-line version, in preserving the ?trouble? of the first line, while eliminating al but the briefest of elaborations, ?with a perfect contempt for it,? functions both as a cutting witicism and a generous invitation. This version simultaneously shuts off avenues for engagement and emphasizes the role of the reader in the act of creating meaning. In this way, ?Poetry? supports not only a first- order analysis of a poetics in which poem and printing history together thematize acts of omision, but also an iluminating second-order analysis, in which poem, revisions, and reception together make clear that the text is indeed a conversation. 192 Sumary An acount of textual discourse that recognizes both the role of individual cognition and the degre to which meaning is an emergent property of interaction is not only interesting as a question of theory. It can also serve to enrich and regulate the practice of historical and editorial encounters with individual texts. The case of ?Poetry? ilustrates thre central findings: 1. A cognitively realistic approach to textual criticism is long overdue. By introducing insights from the study of social cognition and its role in discourse, text, and meaning, editors and archivists can retain the practical methodology of the work/expresion framework while also acknowledging the part that this framework plays in creating a ?social text? of a literary work. 2. The mechanics of joint construal provide a productive framework for talking about these proceses while obscuring neither the degre to which many readers of al stripes sincerely privilege authorial intention nor the social reality in which texts and their meanings are the product of the intersection of many minds. 3. This framework can also open up new readings for and curatorial approaches to individual works. The venerable tradition of identifying in ?Poetry? a poetics in which poem and printing history together thematize acts of omision can now be expanded into an iluminating second-order 193 analysis, in which poem, revisions, and reception together enact a poetics of conversation. The joint construal approach thus provides a number of benefits for the textual analyst. Within the realm of specific textual criticism, it iluminates several aspects of ?Poetry? and its history that are otherwise opaque. First, it gives us a way to acknowledge that the critical narative of omision is both compeling and analyticaly useful, despite being at odds with many apparent facts about the history of the poem. It also sheds new light on how ?Poetry? functions conversationaly, by revealing the poem?s deployment on many levels of the mechanics of revision in conversation. 194 Chapter 7: Conclusion The problem of literary joint atention is the tripartite question of whether shared interpretations of texts exist, where they come from, and how they are achieved. Joint atention is dificult enough in face-to-face interaction; in the case of literature the problems of communication and coordination become more complicated very quickly. I have argued that readers and writers do in fact unconsciously collaborate in constructing joint atentional scenes. These frames of joint atention are often both elaborate and incomplete, and in those elaborations and imperfections much of interest can be found. Appreciating the role that joint atention and social cognition play in literary discourse supplies a framework not only for talking about how texts in general work but also about the problems, kinks, and peculiarities of individual works?and, too, about the particular interactions within individual communities of authors, readers, and editors of those works. It is dificult, in this kind of project, to serve the competing goals of theory and of interpretation. Theory cals for tidy systems that acount neatly for al the mesy diversity of individual artifacts and behaviors. But literature demands to be taken individualy; it is most interesting and valuable in its particulars, and to make global claims about what literature is or how it works is to risk ignoring what makes literature worth caring about. The story I have offered here is an atempt to bridge the general and the specific. 195 This treatment of literary joint atention makes claims about how literary discourse in general can be acommodated in an acount of communication that takes face-to-face communication as basic, foundational, and inescapably grounded in concrete interpersonal interaction. But it is also an exploration of the way that individual texts exploit and thematize these general facts, structured around particular puzzles of joint atention. Literature both appropriates and surpases the limitations of face-to-face interaction. Modernist experimental naratives use joint atention to invite the reader into a character?s consciousnes; detective fiction capitalizes on joint atentional biases to tel stories that mislead and inform the reader in equal measure; poets and their editors direct one another?s atention to particular aspects of a work and its history to create an emergent reading that could not exist without their repeated triangulations. Triangles of joint atention are thus repeated across the literary experience in smal ways and in large, and an understanding of the joint nature of conversation works brings a new richnes to our understanding of texts. 196 Bibliography Albriton, D.W., McKoon, G., and Ratclif, R. 1996. ?The Reliability of Prosodic Cues for Resolving Syntactic Ambiguity.? Journal of Experimental Psychology: Learning, Memory, and Cognition 22: 714-735. Aranda , N. 2007. ?The Pros and Cons of Vinyl Fencing.? EzineArticles. Retrieved August 1, 2007, from htp:/ezinearticles.com/?The-Pros-and-Cons-of-Vinyl- Fencing&id=434398 Ariel, M. 1990. Acesing Noun Phrase Antecedents. London: Routledge. Auerbach, E. 1946 [1968]. W. R. Trask (Trans.) Mimesis: The Representation of Reality in Western literature. Princeton, NJ: Princeton University Pres. Austen, J. 1816 [1999]. Ema. New York: W. W. Norton Bakeman, R. and Adamson, L. 1984. ?Coordinating Atention to People and Objects in Mother-Infant and Per-Infant Interaction.? Child Development 55: 1278- 1289. Baldwin, D.A. 1991. ?Infants' Contributions to the Achievement of Joint Reference. Child Development 62, 875-890. Baldwin, D.A. 1993. ?Early Referential Understanding: Infants? Ability to Recognize Referential Acts for What They Are.? Developmental Psychology 29, 832- 843. Banfield, A. 2000. The Phantom Table: Woolf, Fry, Russel and the Epistemology of Modernism. New York: Cambridge University Pres. 197 Baron-Cohen, S. 1995. Mindblindnes: An Esay on Autism and Theory of Mind. Cambridge, MA: MIT Pres. Barthes, R. 1977. ?The Death of the Author.? Image, Music, Text, trans. S. Heath, New York: Hil and Wang, 142-47. Bateson, M. 1975. ?Mother-Infant Exchanges: The Epigenesis of Conversational Interaction.? In D. Aaronson and R. Rieber (Eds.) Developmental Psycholinguistics and Communication Disorders. Annals of the New York Academy of Sciences, vol. 263 Batiata, M. 2005, May 22. ?Blood Feud.? Washington Post W8. Bigelow, A. E. 2003. ?The Development of Joint Atention in Blind Infants.? Development and Psychopathology 15: 259-275. Birch, S. A. J. 2005. ?When Knowledge is a Curse: Children?s and Adults? Reasoning About Mental States.? Current Directions in Psychological Science 14: 25-29. Birch, S. A. J. and Bloom, P. 2003. ?Children are Cursed: An Asymmetric Bias in Mental-State Atribution.? Psychological Science 14: 283-286. Birch, S. A. J. and Bloom, P. 2004. ?Understanding Children?s and Adults? Limitations in Mental State Reasoning.? Trends in Cognitive Sciences 8: 255- 260. Bloom, P. 1996. ?Intention, History, and Artifact Concepts.? Cognition 60: 1-29. Booth, M. 1997. The Doctor and the Detective: A Biography of Sir Arthur Conan Doyle. New York: St. Martin?s. Booth, W. C. 1961. The Rhetoric of Fiction. Chicago and London: The University of Chicago Pres. 198 Booth, W. C. 1974. The Rhetoric of Irony. Chicago and London: The University of Chicago Pres. Bradbury, M. and McFarlane, J. 1976. ?The Name and Nature of Modernism.? In M. Bradbury and J. McFarlane (Eds.) Modernism 1890 to 1939: A Guide to European Literature 1890-1930. London: Penguin. Bratman, M. E. 1989. ?Intention and Personal Policies.? In J. E. Tomberlin (Ed.) Philosophical Perspectives, Vol. 3: Philosophy of Mind and Action Theory. New York: Blackwel. Bratman, M. E. 1992. ?Shared Cooperative Activity.? Philosophical Review 101: 327-41. Bronshteyn, Y. 2006. ?What We Know Must Be: Hamlet, Religion and Existential Theory of Mind.? Paper presented at the First Literature and Cognitive Science Conference, Storrs, CT. Bruce, B. 1981. ?A Social Interaction Model of Reading.? Discourse Proceses 4: 273-311. Bruner, J. 1977. ?Early Social Interaction and Language Acquisition.? In: H.R. Schafer (Ed.) Studies in Mother-Infant Interaction. New York: Academic Pres. Bruner, J. 1983. Child?s Talk: Learning to Use Language. New York: Norton. Bruner, J. 1995. ?From Joint Atention to the Meting of Minds.? In C. Moore and P. Dunham (Eds.) Joint Atention: Its Origins and Role in Development. Hilsdale, N.J.: Erlbaum. 61-83. 199 Butterworth, G. E. and Jaret, N. 1991. ?What Minds Have in Common Is Space: Spatial Mechanisms Serving Joint Atention in Infancy.? British Journal of Developmental Psychology 9: 55-72. Cal, J., Hare, B., Carpenter, M., and Tomaselo, M. 2004. ??Unwiling? vs. ?Unable?: Chimpanzes? Understanding of Human Intentional Action.? Developmental Science 7:488-498. Camerer, C. F., Loewenstein, G. F., and Weber, M. 1989. ?The Curse of Knowledge in Economic Setings: An Experimental Analysis.? Journal of Marketing 53: 1-20. Carpenter, M., Nagel, K., and Tomaselo, M. 1998. ?Social Cognition, Joint Atention, and Communicative Competence from 9 to 15 Months of Age.? Monographs of the Society for Research in Child Development 63. Chafe, W. 1976. ?Givennes, Contrastivenes, Subjects, Topics, and Point of View.? In C.N. Li (Ed.) Subject and Topic. New York: Academic Pres, 23-55. Chafe, W. L. 1994. Discourse, Consciousnes, and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing. Chicago: University of Chicago Pres. Christie, A. 1926 [2000]. The Murder of Roger Ackroyd. New York: Berkley Books. Clark, H. H. 1996. Using Language. Cambridge: Cambridge University Pres. Clark, H. H. and Marshal, C. R. 1981. ?Definite Reference and Mutual Knowledge.? In A. K. Joshi, B.L. Webber, and I. A. Sag (Eds.) Elements of Discourse Understanding. Cambridge: Cambridge University Pres. 10-63. 200 Clark, H. H. and S.A. Brennan. 1991. ?Grounding in Communication.? In L.B. Resnick, J.M. Levine, and S.D. Teasley (Eds.) Perspectives on Socially Shared Cognition. Washington: APA Books. 127-49. Corkum, V. and Moore, C. 1995. Development of Joint Visual Atention in Infants. In C. Moore and P. Dunham (Eds.) Joint Atention: Its Origins and Role in Development. Hilsdale, NJ: Erlbaum. 61-83. Costelo, B. 1981. Marianne Moore: Imaginary Posesions. Cambridge: Harvard University Pres. Croft, W. 2000. Explaining Language Change: An Evolutionary Approach. London: Longman. Cutrer, M. 1994. Time and Tense in Narrative and in Everyday Language. Ph.D. disertation. University of California, San Diego. Dennet, D. 1987. The Intentional Stance. Cambridge: MIT Pres. Derida, J. 1975 [1988]. ?The Purveyor of Truth.? Trans. Alan Bas. Reprinted in J.P. Muller and W.J. Richardson (Eds.) The Purloined Poe: Lacan, Derida, and Psychoanalytic Reading. Baltimore: Johns Hopkins University Pres, 173- 212. DeSanctis, C. 2004. ?Statutory Interpretation and the A New Approach to ?New Textualism?: A Pragmatic Perspective on A Leading Theory of Statutory Interpretation.? Unpublished MS. DiBatista, M. 1996. ?The Lowly Art of Murder: Modernism and the Case of the Fre Woman.? In M. DiBatista and L. McDiarmid (Eds.) High and Low Moderns: 201 Literature and Culture, 1889-1939. Oxford and New York: Oxford University Pres. 176-196. Dixon, P. and M. Bortolussi. 2001. ?Text Is Not Communication: A Chalenge to a Common Asumption.? Discourse Proceses 31: 1-25. Douglas, M. 2007. Thinking in Circles: An Esay on Ring Composition. New Haven: Yale University Pres. Epley, N. and Kruger, J. 2005. ?When What You Type Isn?t What They Read: The Perseverance of Stereotypes and Expectancies Over Email.? Journal of Experimental Social Psychology 41: 414?422. Epley, N., Keysar, B., Van Boven, L., and Gilovich, T. 2004. ?Perspective Taking as Egocentric Anchoring and Adjustment.? Journal of Personality and Social Psychology 87: 327?339. Eskridge, W. and Frickey, P. P. 1995. Legislation: Statutes and the Creation of Public Policy. St. Paul, MN: West Publishing Eskridge. W. N. 1990. ?The New Textualism.? UCLA Law Review 37: 621-91. Fauconnier, G. 1985. Mental Spaces. New York: Cambridge University Pres. Fauconnier, G. 1997. Mappings in Thought and Language. Cambridge.: Cambridge University Pres. Fauconnier, G. and Turner, M. 2002. The Way We Think. New York: Basic Books. Filmore, C. J. 1982. ?Frame Semantics?. Linguistics in the Morning Calm, ed. Linguistic Society of Korea. Seoul: Hanshin. 111-37. 202 Fischhoff, B. 1975. ?Hindsight Does Not Equal Foresight: The Efect of Outcome Knowledge on Judgment Under Uncertainty.? Journal of Experimental Psychology: Human Perception and Performance 1: 288-299. Flavel, J. H. 1999. ?Cognitive Development: Children?s Knowledge About the Mind.? Annual Review of Psychology 50: 21-43. Flavel, J.H., Gren, F.L., and Flavel, E.R. 1986. ?Development of Knowledge About the Appearance-Reality Distinction.? Monographs of the Society of Research in Child Development 51. Freman, R. A. 1924 [1946]. ?The Art of the Detective Story.? In Haycraft, H. (Ed.), The Art of the Mystery Story: A Collection of Critical Esays. New York: Biblo and Tannen. 7-17. Frost, R. 1915. ?Mending Wal.? North of Boston. New York: Dover Reprint, 28. Galese, V. 2007. ?Before and Below ?Theory of Mind?: Embodied Simulation and the Neural Correlates of Social Cognition.? Philosophical Transactions of the Royal Society B: Biological Sciences 362: 659-669. Galese, V. and Goldman, A. 1998. ?Miror Neurons and the Simulation Theory of Mind-Reading.? Trends in Cognitive Sciences 12: 493-501. Garod, S. and Pickering, M.J. 2004. ?Why Is Conversation So Easy?? TRENDS in Cognitive Sciences 8: 8-11. Gerig, R. J. 1993. Experiencing Narrative Worlds. New Haven: Yale University Pres. Gibbs, R. W. 1998. ?The Varieties of Intentions in Interpersonal Communication.? In S.R. Fussel and R.J. Kreuz (Eds.) Social and Cognitive Psychological 203 Approaches to Interpersonal Communication. Mahwah, NJ: Lawrence Erlbaum, 19-37. Gibbs, R. W. 1999. Intentions in the Experience of Meaning. New York: Cambridge University Pres. Gilovich, T., Kruger, J., and Medvec, V.H. 2002. ?The Spotlight Efect Revisited: Overestimating the Manifest Variability in our Actions and Appearance.? Journal of Experimental Social Psychology 38: 93-99. Gilovich, T., Medvec, V.H., and Savitsky, K. 2000. ?The Spotlight Efect in Social Judgment: An Egocentric Bias in Estimates of the Salience of One?s Own Actions and Appearance.? Journal of Personality and Social Psychology 78: 211-222. Gilovich, T., Savitsky, K., and Medvec, V.H. 1998. ?The Ilusion of Transparency: Biased Asesments of Others? Ability to Read our Emotional States.? Journal of Personality and Social Psychology 75: 332-346. Goffman, E. 1974. Frame Analysis. Boston: Northeastern University Pres. Goffman, E. 1981. ?Footing.? Forms of Talk. Philadelphia: University of Pennsylvania Pres, 124-59. Goldman, A.I. 1992. ?In Defense of the Simulation Theory.? Mind and Language 7: 104-119. Goodwin, C. 1979. ?The Interactive Construction of a Sentence in Natural Conversation.? In G. Psathas (Ed.) Everyday Language: Studies in Ethnomethodology. New York: Irvington Publishers, 97-121. 204 Goodwin, C. 1980. ?Restarts, Pauses, and the Achievement of a State of Mutual Gaze at Turn-Beginning.? Sociological Inquiry 50: 272-302. Goodwin, C. 1986. ?Audience Diversity, Participation and Interpretation.? Text 6: 283-316. Gopnik, A. and Meltzoff, A.N. 1997. Words, Thoughts, and Theories. Cambridge, MA: MIT Pres. Gopnik, A. and Welman, H. 1994. ?The Theory-Theory.? In L. Hirschfeld and S. Gelman (Eds.) Mapping the Mind: Domain Specificity in Cognition and Culture. New York: Cambridge University Pres, 257-293. Graeser, A. C., C. Bowers, B. Olde, and V. Pomeroy. 1999. ?Who Said What? Source Memory for Narator and Character Agents in Literary Short Stories?. Journal of Educational Psychology 91: 284-300. Graf, G. 1987. Profesing Literature: An Institutional History. Chicago: University of Chicago Pres. Gren, J. 2006. ?Theory of Minds: Focalization, Folk Psychology, and Ian McEwan?s Atonement.? Paper presented at the First Literature and Cognitive Science Conference, Storrs, CT. Gren, R. L. 1986. The Sherlock Holmes Leters. London: Secker and Warburg. Gregor, A. 1984. ??Omisions Are Not Acidents?: Reminiscences.? Twentieth Century Literature 30: 150-159. Gregory, E. 1996. Quotation and Modern American Poetry: ?Imaginary Gardens with Real Toads.? Ann Arbor: University of Michigan Pres. 205 Grice, H. P. 1967. ?Logic and Conversation.? Wiliam James Lectures. In H. P. Grice 1989, Studies in the Way of Words. Harvard University Pres, Cambridge MA, 3-143. Haiman, J. 1994. ?Ritualization and the Development of Language.? In W. Paglucia (Ed.) Perspectives of Grammaticalization. Amsterdam: John Benjamins, 3-28. Hare, B. Cal, J., and Tomaselo, M. 2001. ?Do Capuchin Monkeys, Cebus Apela, Know What Conspecifics Do and Do Not Se?? Animal Behaviour 65: 131- 142. Haris, P.L. 1992. ?From Simulation to Folk Psychology: The Case for Development.? Mind and Language 7: 120-144. Hay, D. F. 1979. ?Cooperative Interactions and Sharing Betwen Very Young Children and Their Parents.? Developmental Psychology 15: 647-655. Herman, D. 2006. ?Genete Mets Vygotsky: Narative Embedding and Distributed Inteligence.? Language and Literature 15: 357-380. Hicok, B. 2000. ?To Work ?Lovingly?: Marianne Moore at Bryn Mawr, 1905-1909.? Journal of Modern Literature 23: 483-501. Holley, M. 1984. ?The Model Stanza: The Organic Origin of Moore?s Syllabic Verse.? Twentieth Century Literature 30: 181-191. Honigsblum, B. 1990. ?Marianne Moore?s Revisions of ?Poetry.?? In P. C. Wilis (Ed.) Marianne Moore: Woman and Poet. Orono, ME: National Poetry Foundation, 185-222. Hopper, P. and Traugott, E. C. 2003. Grammaticalization. 2 nd Edition. Cambridge: Cambridge University Pres. 206 IFLA Study Group on the Functional Requirements for Bibliographic Records. 1998. Functional Requirements for Bibliographic Records: Final Report. M?nchen: UBCIM Publications. Irwin, J. T. 1994. The Mystery to a Solution: Poe, Borges, and the Analytic Detective Story. Baltimore: Johns Hopkins University Pres. Israel, M. and Tobin, V. 2006. ?Blending and Irony.? Paper presented at the First Literature and Cognitive Science Conference, Storrs, CT. Jackson, P.L., Meltzoff, A.N., Decety, J. 2005. ?How Do We Perceive the Pain of Others? A Window Into the Neural Proceses Involved in Empathy.? Neuroimage 24 (3): 771-779. Jacobus, M. 1986. Reading Woman: Esays in Feminist Criticism. New York: Columbia University Pres. James, W. 1890 [2007]. The Principles of Psychology, Vol.1. New York: Cosimo. Jeferson, G. 1978. ?Sequential Aspects of Storyteling in Conversation.? In J. Schenkein (Ed.) Studies in the Organization of Conversational Interaction. New York: Academic Pres, 219-48. Johnstone, J. K. 1954. The Bloomsbury Group: A Study of E. M. Forster, Lyton Strachey, Virginia Woolf and Their Circle. New York: Noonday Pres. Joyce, J. 1922 [1986]. Ulyses. New York: Vintage Books Kaminski, J., Reidel, J., Cal, J., and Tomaselo, M. 2005. ?Domestic Goats, Capra Hircus, Follow Gaze Direction and Use Social Cues in an Object Choice Task.? Animal Behaviour 69: 11-18 207 Kappel, A. J. 1991. ?Complete with Omision: The Text of Marianne Moore?s Complete Poems.? In G. Bornstein (Ed.), Representing Modernist Texts: Editing as Interpretation. Ann Arbor: University of Michigan Pres, 125-170. Kenner, H. 1949. ?Baker Stret to Ecles Stret: The Odyssey of a Myth.? The Hudson Review 1(4): 481-499 Kenner, H. 1967. ?Artemis and Harlequin.? National Review 19: 1432-1433. Kenner, H. 1969. ?Meditation and Enactment.? In C. Thomlinson (Ed.), Marianne Moore: A Collection of Critical Esays. Englewood Clifs, N.J.: Prentice Hal, 159-164. Kenner, H. 1975. A Homemade World: The American Modernist Writers. New York: Alfred A. Knopf. Keysar, B., and Bly, B. 1999. ?Swiming Against the Current: Do Idioms Reflect Conceptual Structure?? Journal of Pragmatics 31: 1559-1578. Keysar, B., and Henley, A. S. 2002. ?Speakers? Overestimation of Their Efectivenes.? Psychological Science 13: 207?212. Keysar, B., Bar, D. J., Balin, J. A., and Paek, T. S. 1998. ?Definite Reference and Mutual Knowledge: Proces Models of Common Ground in Comprehension.? Journal of Memory and Language 39: 1-20. Keysar, B., Lin, S., and Bar, D. J. 2003. ?Limits on Theory of Mind Use in Adults.? Cognition 89: 25-41. Keysar, B. 1994. ?The Ilusory Transparency of Intention: Linguistic Perspective Taking in Text.? Cognitive Psychology 26: 165-208. 208 Keysar, B. 2000. ?The Ilusory Transparency of Intention: Does June Understand What Mark Means Because He Means It?? Discourse Proceses 29: 161-172. Knapp, S., and Michaels, W. B. 1982. ?Against Theory.? Critical Inquiry 8: 723-742. Knapp, S., and Michaels, W. B. 1987. ?Against Theory 2: Hermeneutics and Deconstruction.? Critical Inquiry 14: 49-68. Knox, N. 1961. The Word Irony and Its Context, 1500-1755. Durham, NC: Duke University Pres. Krauss, R. M., and Glucksberg, S. 1977. ?Social and Nonsocial Speech.? Scientific American 236: 100-105. Kripke, Saul. 1990. Presupposition and Anaphora: Remarks on the Formulation of the Projection Problem. Unpublished MS. Kruger, J., Epley, N., Parker, J., and Ng, Z. 2005. ?Egocentrism Over Email: Can We Communicate As Wel As We Think?? Journal of Personality and Social Psychology 89: 925-936. Lacan, J. 1972 [1988]. ?Seminar on ?The Purloined Leter.?? Trans. Jefrey Mehlman. Reprinted in J.P. Muller and W.J. Richardson (Eds.) The Purloined Poe: Lacan, Derida, and Psychoanalytic Reading. Baltimore: Johns Hopkins University Pres, 77-82. Lambrecht, K. 1994. Information Structure and Sentence Form: Topic, Focus, and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Pres. Langacker, R. W. 1987. Foundations of Cognitive Grammar: Theoretical Prerequisites. Stanford, CA: Stanford University Pres. 209 Leaky Cauldron. 2000. Transcript: J.K. Rowling at Radio City in ?An Evening with Harry, Carrie and Garp?. Retrieved August 1, 2007, from http:/ww.the- leaky-cauldron.org/page/event-reports/jkrnycnight1 Le, B. P. H. 2001. ?Mutual Knowledge, Background Knowledge and Shared Beliefs: Their Roles in Establishing Common Ground.? Journal of Pragmatics 33(1): 21-44. Lelenberg, J., D. Stashower, and C. Foley. 2007. Arthur Conan Doyle: A Life in Leters. New York: Penguin Pres. Lemire, E. 2001. ??The Murders in the Rue Morgue?: Amalgamation Discourses and the Race Riots of 1838 in Poe?s Philadelphia.? In J. G. Kennedy and L. Weisberg (Eds.) Romancing the Shadow: Poe and Race. New York: Oxford University Pres, 177-204. Leslie, A. and Frith, U. 1988. ?Autistic Children?s Understanding of Seing, Knowing and Believing.? British Journal of Developmental Psychology 6: 315-24. Leslie, A. M., and Polizi, P. 1998. ?Inhibitory Procesing in the False Belief Task: Two Conjectures.? Developmental Science 1: 247?254. Levinson, S. C. 1983. Pragmatics. Cambridge: Cambridge University Pres. Lewis, D. 1979. ?Scorekeeping in a Language Game.? Journal of Philosophical Logic 8: 339-359. Lewis, D. K. 1969. Convention: A Philosophical Study. Cambridge, MA: Harvard University Pres 210 Liszkowski, U., Carpenter, M., Henning, A., Striano, T., and Tomaselo, M. 2004. ?Twelve-Month-Olds Point to Share Atention and Interest.? Developmental Science 7(3): 297-307. Lourdeaux, S. 1982. ?Toads in Gardens for Marianne Moore and Wiliam Carlos Wiliams.? Modern Philology 80, 166-167. MacWhinney, B. 2005. ?The Emergence of Gramar from Perspective.? In D. Pecher and R. A. Zwan (Eds.) The Grounding of Cognition: The Role of Perception and Action in Memory. Cambridge: Cambridge University Pres, 198-223. Mancing, H. 2006. ?Theory of Mind and Literature.? Paper presented at the First Literature and Cognitive Science Conference, Storrs, CT. McGann, J. 1983. Critique of Modern Textual Criticism. Charlottesvile, VA: University Pres of Virginia. McGann, J. 1991. The Textual Condition. Princeton, NJ: Princeton University Pres. McHale, B. 1987. Postmodernist Fiction. London: Methuen. McNeilie, A. 2000. ?Bloomsbury.? In S. Roe and S. Selers (Eds.) The Cambridge Companion to Virginia Woolf. New York: Cambridge University Pres. Meltzoff, A.N. 1995. ?Understanding the Intentions of Others: Re-enactment of Intended Acts by 18-Month-Old Children.? Developmental Psychology 31: 838-850. Miler, C. 1995. Marianne Moore: Questions of Authority. Cambridge: Harvard University Pres. Miler, D. A. 1988. The Novel and the Police. Berkeley and Los Angeles: University of California Pres. 211 Molesworth, C. 1990. Marianne Moore: A Literary Life. New York: Atheneum Monroe, H. and Henderson, A. C. 1932/1934/1938. The New Poetry. New York: Macmilan. Moore, C. and D'Entremont, B. 2001. ?Developmental Changes in Pointing as a Function of Atentional Focus.? Journal of Cognition and Development 2: 109-129. Moore, C. and Dunham, P. (Eds.) 1995. Joint Atention: Its Origins and Role in Development. Hilsdale, NJ: Erlbaum. Moore, M. 1924. Observations. London: Egoist Pres. Reprinted in facsimile in Schultze, R. (Ed.) Becoming Marianne Moore. Moore, M. 1967. Collected Poems. New York: Macmilan and The Viking Pres. Moore, M. 1969. ?Interview ith Donald Hal.? In C. Tomlinson (Ed.) Marianne Moore: A Collection of Critical Esays. Englewood Clifs, NJ: Prentice-Hal. Moore, M. 2005. The Poems of Marianne Moore. G. Schulman (Ed.) New York: Penguin. Morton, A. 1980. Frames of Mind: Constraints on the Common-Sense Conception of the Mental. Oxford: Clarendon Pres. Mundy, P. and Sigman, M. 2006. ?Joint Atention, Social Competence and Developmental Psychopathology.? In D. Cicheti and D. Cohen (Eds.) Developmental Psychopathology, 2 nd ed. Volume One: Theory and Methods. Hoboken, N.J.: Wiley, 7-17. Mundy, P., Sigman, M., Ungerer, J., and Sherman, T. 1986. ?Defining the Social Deficits of Autism: The Contribution of Non-Verbal Communication 212 Measures.? Journal of Child Psychology and Psychiatry and Alied Disciplines 27: 657?669. Narayan, S. 2007. ?Visual and Cognitive Viewpoint in an Interactional Task.? Conference presentation, 10 th International Cognitive Linguistics Conference. Narayan, S. In preparation. Betwen the Boxes: What Comics Can Tel us About Language and Thought. Doctoral disertation, University of California, Berkeley. Nesbit, E. 1899 [2006]. The Story of the Treasure Seekers. San Francisco: Chronicle Books. Newton. E. L. 1990. The Rocky Road from Actions to Intentions. Ph. D. disertation. Stanford University. Nicolson, M. 1929 [1946]. ?The Profesor and the Detective.? In H. Haycraft (Ed.) The Art of the Mystery Story: A Collection of Critical Esays.New York: Biblo and Tannen. 110-128. Nolen, S. 1995. ?Efects of Visible Author on Statistics Texts.? Journal of Educational Psychology 87: 47-65. Onishi, K. H., and Bailargeon, R. 2005. ?Do 15-Month-Old Infants Understand False Beliefs?? Science 308: 255-258. Pascual, E. 2002. Imaginary Trialogues: Conceptual Blending and Fictive Interaction in Criminal Courts. Utrecht: LOT Disertation Series, n. 68. Peterson, J. D. 1990. ?Notes on the Poem ?Poetry?: The Ingenuity of Moore?s Poetic ?Place.?? In P. C. Wilis (Ed.), Marianne Moore: Woman and Poet (pp. 223- 242). Orono, ME: National Poetry Foundation. 213 Pickering, M. and Garod, S. 2004. ?Toward a Mechanistic Psychology of Dialogue.? Behavioral and Brain Sciences 27: 169-226. Poe, E. A. 1841 [1992]. ?The Murders in the Rue Morgue.? The Collected Tales and Poems of Edgar Alan Poe. New York: Modern Library. 141-168. Premack, D. G. and Woodruf, G. 1978. ?Does the Chimpanze Have a Theory of Mind?? Behavioral and Brain Sciences 1: 515-526. Prince, E. F. 1981. ?Toward a Taxonomy of Given/New Information.? In P. Cole (Ed.) Radical Pragmatics. New York: Academic Pres, 223-54. Reddy, V. 2000. ?Coynes in Early Infancy.? Developmental Science 3(2): 186-92. Richter, H. 1970. Virginia Woolf: The Inward Voyage. Princeton: Princeton University Pres. Rosenthal, M. L. 1969. ?Comment: Poets and Critics and Poet-Critics.? Poetry 114, 126. Ross, H. S. and Lollis, S. P. 1987. ?Communication Within Infant Social Games.? Developmental Psychology 23: 241-48. Ryle, G. 1949 [2002]. The Concept of Mind. Chicago: The University of Chicago Pres. Said, E. W. 1975 [1985]. Beginnings: Intention & Method. New York: Columbia University Pres. Sayers, D. 1929 [1946]. ?The Omnibus of Crime.? In Haycraft, H. (Ed.) The Art of the Mystery Story: A Collection of Critical Esays. New York: Biblo and Tannen. 71-109. 214 Schegloff, E. A. 2007. Sequence Organization in Interaction: A Primer in Conversation Analysis. Volume 1. Cambridge: Cambridge University Pres. Schegloff, E. A., Jeferson, G., and Sacks, H. 1977. ?The Preference for Self- Correction in the Organization of Repair in Conversation. Language 53: 361- 382. Scheling, T C. 1960. The Strategy of Conflict. Cambridge, MA: Harvard University Pres. Schober, M. F. and H. H. Clark. 1989. ?Understanding by Addreses and Overhearers.? Cognitive Psychology 21: 211-232. Schulman, G. 1969. ?A Conversation with Marianne Moore.? Quarterly Review of Literature 16: 154-71 Schulman, G. 1986. Marianne Moore: The Poetry of Engagement. Urbana: University of Ilinois Pres. Schulman, G., Ed. 2003. The Poems of Marianne Moore. New York: Penguin. Schulze, R. G., Ed. 2002. Becoming Marianne Moore: The Early Poems, 1907-1924. Berkeley: University of California Pres. Schwartz, M. 1986. ?Coded Sound Paterns, Acrostics, and Anagrams in Zoroaster?s Oral Poetry.? In R. Schmit and P. O, Skjaerv? (Eds.) Studia Grammatica Iranica, Festschrift fur Helmut Humbach. Munich: Verlag, 327-392. Searle, J. 1995. The Construction of Social Reality. New York: Fre Pres. Searle, J. R. 1983. Intentionality, Cambridge, UK: Cambridge University Pres. 215 Simpson, R. C., Briggs, S. L., Ovens, J., and Swales, J. M. 2002. The Michigan Corpus of Academic Spoken English. Ann Arbor, MI: The Regents of the University of Michigan. Sontag, S. 1964. ?Notes on Camp.? In Against Interpretation and Other Esays. New York: Farar, Straus & Giroux. Sperber, D. and Wilson, D. 1986. Relevance: Communication and Cognition. Oxford: Blackwel. Stalnaker, R. 1978. ?Asertion.? Syntax and Semantics 9: 315-332. Stalnaker, R. 2002. ?Common Ground.? Linguistics and Philosophy 25: 701-721. Stein, G. 1937 [1974]. ?Why I Like Detective Stories.? In R. B. Has (Ed.) How Writing is Writen: Volume I of the Previously Uncollected Writings of Gertrude Stein. Los Angeles: Black Sparow Pres. Sterling, B. 1996. ?A Workshop Lexicon.? In R. Wilson (Ed.) Paragons: Twelve Master Science Fiction Writers Ply Their Craft. New York: St. Martin?s Pres, 351-363. Stich, S. and Nichols, S. 1998. ?Theory Theory to the Max: A Critical Notice of Words, Thoughts and Theories by Alison Gopnik and Andrew N. Meltzoff.? Mind and Language 13: 421-449. Stroop, J. R. 1935. ?Studies of Interference in Serial Verbal Reactions.? Journal of Experimental Psychology 12: 643-662. Swales, J. 1990. Genre Analysis: English in Academic and Research Setings. Cambridge: Cambridge University Pres. 216 Talmy, L. 1996. ?Fictive Motion in Language and ?Ception?.? In P. Bloom, M.A. Peterson, L. Nadel, and M.F. Garet (Eds.) Language and Space. Cambridge, MA: MIT Pres, 211-276. Tansele, G. T. 1975. ?Greg?s Theory of Copy-Text and the Editing of American Literature.? Studies in Bibliography 28, 323?60. Taylor, T. J. 1992. Mutual Misunderstanding: Scepticism and the Theorizing of Language and Interpretation. Durham: Duke University Pres. Thomas, R. R. 2000. Detective Fiction and the Rise of Forensic Science. Cambridge: Cambridge University Pres. Tobin, V. 2006. ?Ways of Reading Sherlock Holmes: The Entrenchment of Discourse Blends.? Language and Literature 15: 73-90. Tomaselo, M. 1995. ?Joint Atention as Social Cognition.? In C. Moore and P. Dunham (Eds.) Joint Atention: Its Origins and Role in Development. Mahwah, NJ: Lawrence Erlbaum Asociates. Tomaselo, M. 1999. The Cultural Origins of Human Cognition. Cambridge, MA: Harvard University Pres. Tomaselo, M. 1999. The Cultural Origins of Human Cognition. Cambridge, MA.: Harvard University Pres. Tomaselo, M. 2000. ?First Steps in a Usage Based Theory of Language Acquisition.? Cognitive Linguistics 11: 61-82. Tomaselo, M. 2003. Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge, MA and London: Harvard University Pres. 217 Tomaselo, M. and Akhtar, N. 1995. ?Two-Year-Olds Use Pragmatic Cues to Diferentiate Reference to Objects and Actions.? Cognitive Development 10: 201-224. Tomaselo, M. and Cal, J. 1997. Primate Cognition. Oxford: Oxford University Pres. Tomaselo, M., Carpenter, M., Cal, J, Behne, T. and Moll, H. 2005. ?Understanding and Sharing Intentions: The Origins of Cultural Cognition.? Behavioral and Brain Sciences 28: 675-735. Traugott, E. C. and Dasher, R. B. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Pres. Trevarthen, C. 1979. ?Instincts for Human Understanding and for Cultural Cooperation: Their Development in Infancy.? In M. von Cranach, K. Foppa, W. Lepenies and D. Ploog (Eds.) Human Ethology: Claims and Limits of a New Discipline. Cambridge: Cambridge University Pres. Tuomela, R. 2007. The Philosophy of Sociality. New York: Blackwel. Unfogged.com. 2006, November 16. ?Is Economics Science, Or Are Economists More Like 19th Century Natural Philosophers?? Mesages posted to http:/ww.unfogged.com/archives/comments_5794.html Warden, D.A. 1976. ?The Influence of Context on Children?s use of Identifying Expresions and References.? British Journal of Psychology 67: 101-112. Welman, H. and Gelman, S. 1998. Knowledge Acquisition in Foundation Domains. In D. Kuhn and R. S. Seigler (Eds.) Handbook of Child Psychology, Vol. 2: Cognition, Perception and Language. New York: Wiley, 523-573. 218 Wicker, B., Keysers, C., Plaily, J., Royet, J-P., Galese, V., and Rizolati, G. 2003. ?Both of Us Disgusted in My Insula: The Common Neural Basis of Seing and Feling Disgust.? Neuron 40: 655-664. Wilis, P. C. (Ed.) 1990. Marianne Moore: Woman and Poet. Orono, ME: National Poetry Foundation. Wilis, P.C. 1986. ?Introduction.? In M. Moore, The Complete Prose of Marianne Moore. New York: Viking Pres. Wimer, H. and Perner, J. 1983. ?Beliefs About Beliefs: Representing and Constraining Function of Wrong Beliefs in Young Children?s Understanding of Deception.? Cognition 13: 103-128. Wimsat, W. K., and Beardsley, M. C. 1942. ?Intention.? Dictionary of World Literature, ed. J. T. Shipley. New York: Philosophical Library, 326-9. Wimsat, W. K., and Beardsley, M. C. 1954. ?The Intentional Falacy.? The Verbal Icon: Studies in the Meaning of Poetry. Lexington: University of Kentucky Pres, 3-18 Woodward, A. L. 1998. ?Infants Selectively Encode the Goal Object of an Actor?s Reach.? Cognition 69: 1-34. Woolf, L. 1927. ?The World of Books: Detective Stories.? The Nation & Athenaeum 40: 727. Woolf, L. 1964. Beginning Again: An Autobiography of the Years 1911-1918. London: Hogarth Pres. Woolf, V. 1927 [1981]. To the Lighthouse. New York: Harvest. 219 Wright, W. H. 1927 [1946]. ?The Great Detective Stories.? In Haycraft, H. (Ed.) The Art of the Mystery Story: A Collection of Critical Esays. New York: Biblo and Tannen, 33-70. Zukofsky, L. 1948/1952. A Test of Poetry. New York: The Objectivist Pres. Zunshine, L. 2003. ?Theory of Mind and Experimental Representations of Fictional Consciousnes.? Narrative 11: 270-291. Zunshine, L. 2006. Why We Read Fiction: Theory of Mind and the Novel. Columbus, OH: The University of Ohio Pres.