---------------------------------------------------------------- Sciences of the Text David Herman North Carolina State University dherman@unity.ncsu.edu © 2001 David Herman. All rights reserved. ---------------------------------------------------------------- 1. Sometime between 1966 and 1968, Roland Barthes began to lose faith that there might be a science of the text. This, to be sure, was not an individualized crisis of belief; it was part of a wider transformation at work in the history of literary and cultural theory--in France and elsewhere. Here I shall not try to document, let alone account for, every aspect of this sea-change in theory and criticism.[1] My aims are far more restricted. I mean, first, to conduct a partial genealogical investigation of the notion "science of the text" in Barthes's own discourse. On the basis of this inquiry, I shall then sketch arguments in favor of a research program that should not be dismissed out of hand, without a fair hearing. At issue is an agenda for research that rehabilitates textual science as a legitimate field of endeavor. And--provisionally, at least--I use the word science without scare-quotes. 2. My definition of the term science is, admittedly, a fairly broad one, closer perhaps to the sense of the German word Wissenschaft than to the narrower range of meanings associated with its English cognate. By science I mean the principled investigation of a problem (or set of problems) within a particular domain of inquiry. I acknowledge that "problems," "domains of inquiry," and the principles according to which investigation can be "principled" are historico-institutional constructs, not necessarily reflections of the way things truly are. That said, the present essay attempts to reinflect social-constructionist arguments pursued by proponents of the social study of science, for example. Scholars such as Malcolm Ashmore, David Bloor, and Steve Woolgar have demonstrated that scientific practice is always embedded in a particular social context. What counts as scientific, and more specifically what marks the border between "scientific" and "nonscientific" (e.g., humanistic) modes of inquiry, is historically variable. Thus, for Woolgar, "there is no essential difference between science and other forms of knowledge production" (Science 12). Rather, scholars must now "accept that science cannot be distinguished from non-science by decision rules. Judgements about whether or not hypotheses have been verified (or falsified), as to what constitutes the core or periphery in a research programme, and at what point to abandon a research programme altogether, are the upshot of complex social processes within a particular environment" (17). In this way "the ethnographic study of science... portrays the production of scientific facts as a local, contingent accomplishment specific to the culture of the laboratory setting" ("Reflexivity" 18). 3. But by the same logic, the boundary between humanistic and (social-)scientific research should be viewed not as fixed and impermeable but rather as shifting and porous. My essay centers around a particular instance of this general proposition, examining how the structuralist method articulated by the early Barthes involved an attempt to redraw the border between the science of language and the theory of literature. That attempt can now be reevaluated in light of more recent research in discourse analysis, the field of linguistics that studies units of language larger than the sentence. There were, it is true, important precedents for the structuralists' efforts to span the disciplinary divide between linguistics and literature--a divide that might be better characterized as an unstable seam in the architecture of inquiry. For example, whereas Ferdinand de Saussure distrusted written data as a basis for the structural analysis of language (23-32), the great speculative grammarians of the Middle Ages used literary language to develop theories about the homology between vox (words), mens (mind), and res (things) (Herman, Universal Grammar 7-14). In contrast to the speculative grammarians of the late thirteenth and early fourteenth centuries, however, twentieth-century structuralists like the early Barthes were adapting linguistic methods and ideas at the very moment when language theory was itself undergoing revolutionary changes. 4. Those changes stemmed, in part, from emergent formal (e.g., generative-grammatical) models for analyzing language structure (cf. Chomsky's 1957 and 1964 publications, Syntactic Structures and Aspects of the Theory of Syntax). But the changes also derived from an increasing concern with how contexts of language use bear crucially on the production and interpretation of socially situated utterances--as opposed to the decontextualized sentences (or "sentoids") that are still the staple of many linguistics textbooks. In the first instance, even as Barthes and Claude Lévi-Strauss were drawing on the linguistic structuralism of Saussure, Roman Jakobson, and Louis Hjelmslev to write texts such as Elements of Semiology and Structural Anthropology, linguistic science was moving from the Saussurean-Hjelmslevian conception of language as a system of similarities and differences to a Chomskyean conception of language as a "discrete combinatorial system" whereby "a finite number of discrete elements [e.g., words] are sampled, combined, and permuted to create larger structures [e.g., sentences] with properties that are quite distinct from those of their elements" (Pinker 84). New, quasi-mathematical formalisms were required to model the workings of this recursively organized system, which has the capacity to operate on its own output and thereby produce such complex strings as The house that the family built that stood on the shoreline that was eroded by the storm that originated from a region that.... At the same time, language theorists working on a different front began to question what they viewed as counterproductive modes of abstraction and idealization in both structuralist linguistics and the Chomskyean paradigm that displaced it. From this other perspective, generative grammarians had perpetuated Saussure's foregrounding of la langue over la parole, language structure over language use, by taking as their explicandum linguistic competence and jettisoning a host of phenomena (conversational disfluencies, nonliteral usages, differences in speech styles, etc.) that generative grammarians viewed as ignorable--i.e., as matters of linguistic performance only. 5. For example, Chomsky sought to characterize linguistic competence by abstracting away from the complexities of real-world communication and identifying the cognitive skills and dispositions needed for an idealized speaker and hearer (using a homogenous communicative code) to produce and understand mutually intelligible sentences. By contrast, sociolinguists ranging from Erving Goffman and Dell Hymes to William Labov and J. J. Gumperz insisted on the inherent variability of communicative codes: anyone's use of any language on any occasion is, in effect, a particular variety or dialect of that language. Speakers can furthermore choose between more or less formal speech styles, as well as registers more or less appropriate for a given situation. We can choose, for instance, between I wonder, may I please have a beer? and Give me a beer; we can also select from among various lexical and discourse options that will be more or less suitable for conversing with a coworker, engaging in a service encounter at the local convenience store, or composing an academic essay. In addition, considerations of age, social status, gender identity, and so on constrain which communicative options potentially available to interlocutors using a given code are typically or preferentially selected by them. Hence, from the perspective of contextually oriented theories of language, formalist idealizations of speakers and hearers are methodologically suspect, yielding models of linguistic competence (i.e., competence at producing and understanding grammatically well-formed strings) that need to be supplanted by ethnographically based models of communicative competence (i.e., competence at producing and interpreting different sorts of utterances in different sorts of situated communicative events) (Hymes 3-66; Saville-Troike 107-80, 220-53). Linguistic competence accounts for my ability to produce the string Look, that person over there is wearing the worst-looking coat I've ever seen!; communicative competence accounts for my tendency to refrain from actually producing the string in question, in all but a very few communicative circumstances. 6. As even this thumbnail sketch suggests, within the field of linguistics itself there are significant disagreements over what properly constitutes the data, methods, and explanatory aims of the science of language, with many of the disagreements at issue starting to crystallize around the time that Francophone structuralists began outlining their project for a science of the text. Yet because of the interests and aptitudes of the commentators who have concerned themselves with structuralism's legacy, structuralist notions of textual science have remained, for the most part, dissociated from neighboring developments in linguistic analysis.[2] My purpose here is thus to reassess the problems and potentials of the structuralist project by reattaching it to a broader context of language-theoretical research. Although structuralist methods have been dismissed as misguided, self-deluded, or worse, I reject the dominant characterization of structuralism as a futile exercise in hyper-rationality, a destructive rage for order. I also dispute the orthodox view that structuralist literary theorists such as Barthes were engaged in a doomed attempt to scientize (the study of) literary art. Instead, I contend that Barthes and his fellow-travellers made a productive, consequential effort to reconfigure the relationship between critico-theoretical and linguistic analysis--to redraw the map that had, in the years preceding the rise of structuralism, fixed the positions of humanistic and scientific inquiry in cognitive and cultural space. In the mid-twentieth century, granted, neither literary theory nor linguistics had reached a stage at which the proposed reconfiguration could be accomplished. There is thus a sense in which the structuralist revolution envisioned by the early Barthes (among others) has started to become possible only now. 7. Taken as a case-study, Barthes's emerging ideas about methods for textual analysis reveal larger problems with the way the history of critical theory is sometimes written. In particular, the widespread tendency to chronicle structuralist thought as a brief unfortunate episode of scientism needs to be reconsidered--along with the notion of "scientism" itself. One of my guiding assumptions is that Barthes prematurely stopped using the concept "science of the text" as what Kant would have called a regulative ideal, a goal that orients thought and conduct (Kant 210-11, sections A179-180/B222-223). The project of developing a principled, linguistically informed approach to textual analysis has arguably taken on even greater urgency in the years since Barthes, and those influenced by him, ceased to work in this area. To put things somewhat more colloquially, when at a certain point they abandoned the attempt to use linguistic models to articulate a science of the text, Barthes and his cohorts threw out the baby with the bathwater. After all, the inadequacy of Saussurean models for textual analysis in no way impugns the original insight of the structuralists: namely, that language theory provides invaluable resources for analyzing literary discourse. Post-Saussurean developments in linguistics--specifically, developments in the burgeoning field of discourse analysis--can yield productive new research strategies for analyzing texts. Taken together these new strategies constitute a revitalized textual science, or rather a field of study that needs to be defined by complementary sciences of the text. 8. Outlining prolegomena for new sciences of the text, the second section of my essay contains a brief illustrative analysis of a scene drawn from Virginia Woolf's To the Lighthouse.[3] I focus on literary discourse as a way of putting the rehabilitation of textual science to its severest test. A fundamental question is this: To what extent can research models associated with the science of language help illuminate the language used in literary art, especially when it comes to literary works that (more or less reflexively and ludically) focus on the nature and functions of language itself? Woolf's text foregrounds dimensions of language structure and use not describable, let alone explainable, in structuralist terms. Debatably, however, structuralist approaches to literary analysis were problematic not because they aspired to the status of science, but because they mistook what any such science would have to look like. Specifically, the sciences of the text must be integrative rather than immanent; intuitions about textual structures and functions depend not just on competence at selecting and combining formal units, but on the broad communicative and interactional competence both displayed and created by discourse events, including those events recorded in the form of literary texts. Hence textual analysis will become more principled in proportion to its ability to synthesize a variety of research models to study the linguistic, interactional, cultural, and cognitive skills by virtue of which textual patterns are built up, recognized, and used for any number of communicative purposes. My examination of the scene from Woolf's novel in section II gestures toward a synthesis of this sort, sketching out what will be entailed by redesigning, rather than rejecting, the sciences of the text. I. Barthes's Bouquet: Why a Text Is More than the Sum of Its Sentences 9. What might be called Barthes's early methodological utopianism, his confidence in the possibility of extending Saussurean language theory across broad domains of linguistic and cultural activity, reached its apex in essays such as "The Structuralist Activity," published in 1964, and "Introduction to the Structural Analysis of Narratives," published in 1966. In the former essay, Barthes identified signs of the structuralist activity in the work not only of Troubetskoy, Propp, and Lévi-Strauss, but also of Mondrian, Boulez, and Butor. Both analysts and artists are engaged in the same enterprise--the articulation of "a certain object... by the controlled manifestation of certain units and certain associations of these units"--and it matters little "whether [the] initial object is drawn from a social reality or an imaginary reality" (1197). At this heady stage in Barthes's thinking, one might say, not just the analysis but the production of text is a science of the sort envisioned by Saussure. By the same token, in his "Introduction to the Structural Analysis of Narratives," Barthes drew on Saussure's distinction between la langue and la parole in attempts to find what he called "a principle of classification and a central focus for description from the apparent confusion of individual [narrative] messages" (80). No utterance would be intelligible (or even possible) in the absence of an underlying system of contrastive and combinatory relationships built into the structure of the language in which the utterance is couched. Similarly, argues Barthes, it would be impossible to produce or understand a narrative "without reference to an implicit system of units and rules" (81). As one of the possible object-languages studied by what Barthes describes as a secondary linguistics--i.e., a linguistics not of sentences but of discourses (83)--narrative texts can be construed as higher-order messages whose langue it is the task of structuralist analysis to decode. Though a science of narrative discourse may not yet have been realized in fact, there was for Barthes at this point nothing to indicate that a science of the narrative text could not be accomplished in principle. 10. By the time he published "The Death of the Author" in 1968, however, Barthes had begun to speak about literary discourse in a very different way.[4] Resisting the use of words like code and message as terms of art, and reconceiving texts as gestures of inscription rather than vehicles for communication and expression (146), Barthes had come to embrace a Derridean view of the text as "a tissue of signs, an imitation that is lost, infinitely deferred" (147). The text is, as Barthes now put it, "not a line of words releasing a single 'theological' meaning... but a multi-dimensional space in which a variety of writings, none of them original, blend and clash" (146). The scientific decoding of messages has given way to the interpretative disentanglement of strands of meaning--strands more or less densely woven together by the scriptor who fabricates, but does not invent, the discourse. In "From Work to Text," published in 1971, Barthes characterized the irreducible plurality of the text in similar terms, writing about "the stereographic plurality of its weave of signifiers (etymologically, the text is a tissue, a woven fabric)" (159). To quote another portion of this same passage: [The text] can be it only in its difference (which does not mean its individuality), its reading is semelfactive (this rendering illusory any inductive-deductive science of texts--no "grammar" of the text) and nevertheless woven entirely with citations, references, echoes, cultural languages... antecedent or contemporary, which cut across it through and through in a vast stereophony. (159-60) Barthes here disavows the possibility of a science that just a few years earlier he had, if not taken for granted, then assumed as the outcome toward which structuralist research was inexorably advancing. 11. A moment ago I alluded to Jacques Derrida's writings as a factor influencing the methodological (or metatheoretical) shift that can be detected in Barthes's comments on textual analysis; this shift manifests itself in the movement from terms like unit, articulation, classification, rule, and system, to terms such as tissue, fabric, clash, echo, and stereophony. Derrida's role in the rethinking and radicalization of structuralist semiology has already been well-documented (cf. Dosse). Barthes's shift from the notion of "expression" to the idea of "inscription," for example, clearly bears the impress of Derrida's critique of what he called the transcendental signified (Derrida, "Structure" 83-6; Grammatology 44-73). Less attention, however, has been devoted to the way other, intratextual factors--factors pertaining to Barthes's original formulation of the nature and scope of textual science--may also have motivated the change in question. Indeed, these other factors do much to account for Barthes's (and others') susceptibility to the influence of Derrida's views about signs, meanings, and texts. Arguably, Barthes was eventually driven to deny the possibility of a science of the text because, in his early work, he lacked the resources to identify key structural properties of texts. He was ipso facto unable to model how such structural properties bear on the design and interpretation of discourse. By contrast, in the years since the heyday of structuralism, linguistic research on extended discourse has demonstrated that certain features and properties of language emerge only at the level beyond the sentence. Researchers have also developed powerful new theories for studying ways in which language-users rely on these discourse-level features and properties to negotiate meanings, to build models of the world, to encode information about temporality, spatial proximity, and relative social status--in short, to communicate in the broadest sense of that term. 12. Some of the relevant issues may come into better focus through a reexamination of Barthes's "Introduction to the Structural Analysis of Narratives," and more particularly of the pages where Barthes discusses the relationship between sentences and discourse (82-84) as a prelude to his account of stories as just "one... of the idioms apt for consideration by the linguistics of discourse" (84). Here Barthes draws on the work of André Martinet in arguing that "there can be no question of linguistics setting itself an object superior to the sentence, since beyond the sentence are only more sentences--having described the flower, the botanist is not to get involved in describing the bouquet" (82-83). But the analogy does not really hold up. Discourse context confers linguistically describable properties on utterances (and parts of utterances) that would not have such properties in isolation from the context in which they occur.[5] 13. For example, in the midst of telling someone else about the present essay, one of the readers of this journal might use a definite description like the boring essay to refer to my article. Viewed in isolation, the noun phrase the boring essay has no identifiable feature that would mark it as (part of) an utterance that functions anaphorically. Yet when used in an extended discourse about good and bad journal articles and the differences between them, this definite description could very well pick out an entity previously mentioned in the discourse--namely, this boring essay versus Jones's lively essay or Wasowski's controversial one.[6] The same argument, of course, can be extended from noun-phrases to full clauses and sentences. Think of all the discourse functions that might possibly attach to a sentence like It's raining. Depending on the occasion of talk in which it is issued as an utterance, this sentence might be an indirect speech act that functions as a request for someone to close a window (see section II below); a description of climatic conditions contemporaneous with the time of speaking; a narrative proposition describing a past event but couched in the historical present tense; or a predictive utterance made by a blindfolded prisoner during an interrogation in a windowless room. 14. My larger point here is that research done in the fields of linguistic pragmatics and discourse analysis over the past couple of decades suggests serious problems with the way Barthes defined discourses or texts as mere agglomerations of sentences. There are, in other words, significant grounds for rejecting what Barthes described as a postulate of homology between sentence and discourse. As Barthes put it in his "Introduction," [we can] posit a homological relation between sentence and discourse insofar as it is likely that a similar formal organization orders all semiotic systems, whatever their substances and dimensions. A discourse is a long "sentence" (the units of which are not necessarily sentences), just as a sentence, allowing for certain specifications, is a short "discourse." (83) As already noted, however, a discourse is not a long sentence. Texts mean in a way that is not strictly componential; in contrast to your knowledge about sentence meaning, you do not necessarily know the meaning of a text if you know the meaning of its parts and the relations, logically specifiable, into which those parts can enter. Rather, during the process of textual interpretation, types of communicative competence come into play that are broader than the linguistic competence based on knowledge about truth conditions for sentences and dependency relations between sentence elements. Because the linguistic models on which the early Barthes relied underspecified discourse structure, making it basically tantamount to sentence structure, Barthes quickly exhausted the descriptive and explanatory yield of those models for the purposes of textual analysis. The models did not furnish an adequate definition of what a text is. Nor did they account satisfactorily for the skills required to fashion and understand texts as ways of communicating. In turn, lacking more nuanced theories of textual structure, yet still eager to draw on the semiological revolution as a resource for textual analysis, Barthes shifted from describing texts as instruments of expression to characterizing them as gestures of inscription. Signs, now, deferred signifieds dilatorily. Discourse was severed from communication, text-interpretation from the framing of inferences about a speaker's or scriptor's beliefs. Overall, there seemed to be no good reason to try to anchor texts in language-users' models for understanding the world, their norms for interaction, or their tacit knowledge of the speech events in which particular speech acts are embedded. 15. By rejecting the initial reduction of texts to mere collocations of sentences, however, one can avoid heading down the path that leads to a view of texts as things or events that inscribe without expressing, signify without communicating. And by not taking that path, one arguably remains more "faithful" to the structuralist ideal of textual science than were the structuralists themselves. In this spirit, I should like to turn now from a consideration of what was and might have been to some remarks about what may yet be--provided that we restore to the notion "science of the text" its former (if short-lived) status as a regulative ideal for research. Using a scene from Woolf's To the Lighthouse as my tutor text, I shall discuss features of the text irreducible to features associated with sentence-structure and -meaning. These higher-level features include the following: (1) ways in which the text encodes a complex relation between locutionary and illocutionary acts, or acts of saying and acts of meaning; (2) the manner in which those speech acts are embedded in an overarching speech event, subject to ethnographic description; and (3) strategies by which Woolf at once creates and portrays what Goffman would characterize as a participation framework, in terms of which the participants in the discourse align themselves with one another in certain ways, shift their footing, then take up new alignments. This is of course only a partial inventory of relevant discourse features. My purpose is not to attempt an exhaustive description of all salient properties of the scene, but to use it to promote further debate about the possibilities and limits of textual science. II. Literary Dialogue in a Discourse-Analytic Context 16. I should preface my sample analysis with a methodological proviso. The discourse-analytic models on which I am drawing--models originating in speech act theory, interactional sociolinguistics, and the ethnography of communication--were not designed to account for fictional representations of discourse such as we find in Woolf's novel.[7] To mention just one potential problem, fictional representations such as Woolf's arguably cannot encode all of the "illocutionary force indicating devices" that have been described by speech-act theorists and that are available to discourse participants in other kinds of communicative settings (Searle 30-33). Such devices include, for example, intonational contours, pitch, loudness, and a variety of paralinguistic cues like pauses, conversational synchrony (or asynchrony), head movements, and bodily orientation.[8] I assume in what follows that this methodological difficulty, although significant, will not prove fatal to the refashioning of textual science after Barthes. 17. The scene under discussion occurs in the last chapter (chapter 19) of "The Window," the first section of To the Lighthouse (Woolf 117-24); it is thus placed just before the "Time Passes" section that serves as a bridge between the first and third parts of the novel, and that records the devastating losses experienced by the Ramsay family during the First World War. The scene in question is hence the last scene in which Mrs. Ramsay is still living. 18. Consider, first, the way the text foregrounds the complexity of the relation between locutionary acts--acts of saying, whereby one issues an utterance--and illocutionary acts--acts performed on the basis of the act of saying, whereby one does or means something by the issuing of the utterance. More precisely, the text features types of utterance that would be categorized as "indirect speech acts" in the canonical version of speech act theory outlined by J. L. Austin and then systematized by John Searle. In utterances of this sort, illocutionary force disagrees with surface form--as when I utter Can you pass me the salt? in a context in which I want my interlocutor to pass me the salt, not provide an account of his or her reaching, grasping, and passing abilities. In Woolf's text, there are in fact few true speech acts jointly elaborated by Mr. and Mrs. Ramsay. If we discount the narrator's reports of thoughts that are not outwardly expressed but rather internally verbalized--instances of what Dorrit Cohn would call "psychonarration"--there are just nine speech acts in the entire scene, seven by Mrs. Ramsay ("'Well?'"; "'They're engaged,... Paul and Minta'"; "'How nice it would be to marry a man with a wash-leather bag for his watch'"; "'No,... I shan't finish it'"; "'Yes, you were right. It's going to be wet tomorrow. You won't be able to go'") and two by Mr. Ramsay ("'So I guessed'"; "'You won't finish that stocking tonight'"). (Mrs. Ramsay also quotes a line of poetry sotto voce [122].) But as the surrounding narratorial commentary suggests (e.g., "'Well?' she said, echoing his smile dreamily, looking up from her book" [122, my emphasis]), these skeletal locutions carry complicated, highly nuanced illocutionary forces. Hence Mr. Ramsay's acerbic "'You won't finish that stocking tonight'" (123) functions not just to predict a state of affairs but also simultaneously to badger, reassure, and comfort Mrs. Ramsay, while likewise serving as an invitation to her to tell Mr. Ramsay that she loves him (123). Indeed, Woolf's mode of narration compels us to rethink what the notions "literal" and "indirect" might mean in connection with speech acts.[9] Indirectness may be a quite loose way of talking about talking, given that, at the very end of this scene, Mrs. Ramsay feels that she has in effect been able to "tell" her husband something without "saying" it at all (124). 19. Indeed, as Stephen C. Levinson points out, indirect speech acts of the sort represented in Woolf's text present a challenge to the "literal force hypothesis" associated with classical speech act theory (263-78). According to this hypothesis, "illocutionary force is built into sentence form" (263), such that imperative, interrogative, and declarative sentences have the forces traditionally associated with them, i.e., ordering (or requesting), questioning, and stating, respectively. As already noted, though, in an utterance like Can you pass me the salt?, there is a mismatch between surface form (interrogative) and illocutionary force (ordering/requesting). Woolf portrays Mr. and Mrs. Ramsay's interchange as one consisting almost entirely of such "nonliteral" utterances. The scene thus bears out Levinson's contention that most usages are indirect in this sense (264)--especially when it comes to imperatives that tend to be perceived as a threat to recipients' negative face wants, their desire not to be imposed upon by others (Brown and Levinson). More than this, and as Woolf's text suggests, there are indefinitely many ways of mitigating a request through indirection. A request for someone to pass the salt might be couched in any number of surface forms, such as My, but this food is underseasoned! or I wonder how this food might be made less bland? or How I wish I were sitting closer to the salt shaker or You seem to be really skilled at passing people salt or May the gods rain down salt upon my food. 20. In other words, the human proclivity for indirectness provides a compelling reason for not trying to separate out a level of illocutionary force built into--or at least prototypically associated with--particular sentence forms (Levinson 283). Analogously, Woolf's text underscores the potentially wide disparity between locutionary and illocutionary acts, between modes of saying and strategies for meaning. To rephrase this last point, the scene from To the Lighthouse reveals what might be characterized as the multifunctionality of acts of saying. At stake are both a one-many and a many-one relationship: a single locutionary act can carry multiple illocutionary forces, while a given illocutionary force can be realized by any number of locutionary acts (or even by no locutionary act whatsover). Yet the difficulty of correlating forms with forces and forces with forms does not by and large derail people's attempts to communicate. It is simply not the case that illocutionary forces pattern randomly with utterance form. This suggests that, in modeling language-users' ability to understand what speakers mean on the basis of what they say, researchers should complement bottom-up with top-down approaches to discourse comprehension. Theorists should not only work their way up from analyzing individual speech acts to the way they are sequenced in larger stretches of text; they should also work their way down from (strategies for) text interpretation to analysis of the speech acts embedded in discourse. Or, to paraphrase Levinson, the problem of indirect speech acts suggests the advantages of complementing microanalysis of locutionary forms with a more macroanalytic inquiry into communicative intention, utterance function, and interactional context. All of these factors fall under the purview of the new sciences of the text, which focus on how discourse participants (including fictional characters) use language to get meanings across. 21. Here the usefulness of ethnographic models for textual science makes itself felt. Ethnographers of communication, drawing on Dell Hymes's SPEAKING grid, have located utterances within a nested structure of speech situations, speech events, and speech acts.[10] (Think here of the differences between a colloquium, a lecture given by one of the participants, and then an illocutionary act, e.g., an assertion or a request, occurring within that person's talk.) Mnemonically associated with each letter in the grid, the factors of Setting, Participants, Ends, Act-sequence, Key, Instrumentalities, Norms, and Genre collectively define a speech situation. The way the factors are realized in a given communicative encounter determines what kind of speech event is involved. Thus, participants relate to one another (and the setting) differently in an interview than they do in a conversation, applying different interpretative norms to the speech act sequences that constitute the event, being more or less restricted as to variations in key and genre, and having different ends in view in each case. 22. Along the same lines, at the beginning of chapter 19, Woolf specifies the setting and also the participants: Mr. Ramsay is reading, and Mrs. Ramsay is reading and knitting, in their reading room. The key fluctuates from the pathetic to the humorous to the flatly descriptive. Further, partly because of what Mr. and Mrs. Ramsay are reading, the discourse vacillates between different genres, from poetry, to romance, to commentary, to banter. More significant, Woolf represents these characters as participating in a communicative event that they jointly elaborate as a conversation. Note that they enjoy considerable latitude in co-constructing an act sequence that is not strictly dovetailed with the accomplishment of a particular task. Note, too, that the interpretative norms guiding their speech productions reflect the comparatively fuzzy contours of an event that need not unfold in any particular way. And Mr. and Mrs. Ramsay each have their own, more or less covert, ends in performing the acts of saying that they do in fact perform. 23. In representing this communicative encounter, then, Woolf anchors her text in the same constellation of factors that bears on speech situations at large. Taken together these factors determine what speech event is transpiring from the participants' standpoint. Mr. and Mrs. Ramsay's emergent understanding of the discourse as a certain type of event is what enables them to interpret particular acts of saying as part of a more global textual structure. At another level, ongoing assessments about event-type also allow the reader to interpret the act sequence. Structuralist attempts to articulate a science of the text did not focus enough attention on such typicality judgments or their role in discourse interpretation. 24. Work stemming from interactional sociolinguistics provides additional insights in this connection; this work, too, suggests the advantages of an integrative, holistic approach to textual structure and meaning. Sociolinguists such as Gumperz and Goffman have developed theoretical resources for studying how discourse participants work to interpret sequences of utterances by contextualizing them--in the double sense of situating those utterances in a particular context and specifying what sort of context they mean for the ongoing discourse to create. For example, Goffman has used the notion of "participation frameworks" to rethink older, dyadic models of communication, based on the speaker-hearer pair (124-59). The terms speaker and hearer, on this view, are insufficiently nuanced to capture the many (and fluctuating) statuses that one can have as a discourse participant. These statuses include, on the one hand, speaking as an author, animator, principal, or figure, and, on the other hand, listening as an addressee, an unaddressed but ratified participant, or an unaddressed and unratified participant--e.g., an eavesdropper or a bystander. Further, participants constantly change their "footing" in discourse, thereby changing "the alignment [they] take up to [themselves] and the others present as expressed in the way [they] manage the production or reception of an utterance" (128). 25. In the scene from To the Lighthouse, both Mr. and Mrs. Ramsay display one sort of footing when they animate utterances authored by other people, i.e., the writers whose works they are reading. (For the most part, they "animate" these utterances only internally, by way of psychonarration, though Mrs. Ramsay does murmur the line of poetry [122].) In effect, the characters' acts of animation model the reader's own animating acts: the text encodes the participation framework in terms of which its own analysis is designed to unfold. At the same time, throughout the scene, we readers are unaddressed but ratified participants in the communicative encounter taking place in this fictional world. By making readers participants in the scene it cues them to animate, the text inserts readers in a certain way in the action sequence for which it provides a kind of verbal blueprint. Interpreters' ability to understand Woolf's text as a text hinges on their ability to use this blueprint to reconstruct the scene as a coherent whole--a whole whose coherence derives in part from their own specified mode of participation in it. Within the scene itself, Mr. and Mrs. Ramsay change their footing when they begin to converse. They are now not just animators but also authors of the words they animate. Arguably, however, Mr. Ramsay is the principal for whose sake both Mr. and Mrs. Ramsay speak.[11] Mrs. Ramsay designs her utterances to accommodate her husband's need for external affirmation, whereas Mr. Ramsay does not display the same concern with tailoring his speech to the needs of his wife. Mrs. Ramsay is the one who opens the exchange with a question inviting her husband to speak, and all of her turns at talk are designed, in one way or another, to elicit his opinion. Indeed, Woolf herself provides a striking metaphor for this kind of alignment between participants--a mode of alignment rich with implications for the study of discourse and gender--when she describes how Mrs. Ramsay felt her husband's "mind like a raised hand shadowing her mind" (123). 26. Thus, from the vantage-point I have started to sketch here, a perspective afforded by integrating ideas from speech act theory, the ethnography of communication, and interactional sociolinguistics, acts of saying take on textual functions because of the way discourse itself is situated in, and helps constitute, sociointeractional contexts. More generally, in the textual science that structuralist theorists glimpsed but could not fully articulate, research begins with the recognition that while a text corresponds to a bounded, integral event, its boundaries are negotiated by participants and its units are saturated with social meanings. It consists not just of words and sentences but of verbal acts made intelligible by what readers and interlocutors know about language, other people, and the world. III. Textual Science and Literary Theory 27. I suggested earlier that the structuralist legacy may be, not the lingering trace of a scientism that analysts should work to extirpate from literary theory, but rather the impetus for reassessing what constitutes a principled approach to the analysis of literary as well as nonliterary texts. Structuralism provides this impetus both positively and negatively: it stands as an important precedent for rethinking the scope and aims of literary theory; but it also indicates how the sciences of the text should not be articulated. In its first phase, the project of textual science, true to its Saussurean heritage, went too far in subsuming literary parole under literary langue. It sought to reduce individual messages to the communicative codes by which such messages are produced and interpreted. Misconstruing texts or discourses as agglomerations of sentences, the structuralists also misconstrued the nature of the codes on which they themselves placed so much emphasis. More recent developments in discourse analysis suggest that texts have meaning by virtue of the relationships among messages, codes, and contexts. More precisely, messages or individual texts are made possible by a code that builds in information, first, about the way the parts of a text relate to one another; and second, about the way the text and its parts relate to a particular context of use. The structuralists were thus unable to ask, let alone answer, questions that form the heart of the approach to textual science outlined here--e.g., how a given textual segment evokes an entity referred to in a previous textual segment, and how a stretch of text anchors itself to some essential point in the surrounding context, whether it be the social identities of the interlocutors involved, the overarching communicative event in which the text is embedded, or the participation framework structuring the speech exchanges represented textually. 28. Insofar as these questions can be asked about all sorts of texts, literary as well as nonliterary, they encourage researchers to explore how aspects of literary interpretation bear on the larger enterprise of developing models for the analysis of discourse in general. Yet this is precisely what Barthes said working out the notion of "text" would entail. To quote some characteristically elegant phrasing from "From Work to Text": The work is a fragment of substance, occupying a part of the space of books (in a library for example), the Text is a methodological field.... The work can be seen (in bookshops, in catalogues, in exam syllabuses), the text is a process of demonstration.... The Text is not the decomposition of the work, it is the work that is the imaginary tail of the Text; or again, the Text is experienced only in an activity of production. It follows that the Text cannot stop (for example on a library shelf); its constitutive movement is that of cutting across (in particular, it can cut across the work, several works).... What constitutes the Text is... its subversive force in respect of the old classifications. (156-57) Barthes's remarks were perhaps more future-thinking than he knew. Debatably, it is only now, with the help of linguistic and discourse-analytic concepts to which the structuralists did not have access, that researchers can start reconceiving the literary work as a mode of situated textual practice. In other words, literary texts are definable as discourse productions/events that can be assigned coordinates within an overarching system of sociointeractional parameters. Other sorts of texts--VCR instructions, legal briefs, political speeches, conversational narratives--can be assigned different coordinates within that same sociocommunicative system. Textual science has now begun to amass the tools needed both to characterize the system as a whole and also to pinpoint where particular types of texts are located in the system.[12] Participation frameworks function differently in contexts of literary interpretation than in Supreme Court hearings; these two speech situations entail, as well, very different judgments about what kinds of verbal exchanges are typical or preferred, how many (and what sorts of) indirect speech acts are allowable, and so on. But both kinds of discourse events fall within the purview of textual science, whose widened, cross-disciplinary scope is what will enable it to capture the specificity of literary as opposed to nonliterary discourse. 29. In this way, the new sciences of the text can help bring about the radical interdisciplinarity to which structuralism aspired but which it ultimately failed to achieve. This is the sort of interdisciplinarity that happens "when the solidarity of the old disciplines breaks down... in the interests of a new object and a new language neither of which has a place in the field of the sciences that were to be brought peacefully together, this unease in classification being precisely the point from which it is possible to diagnose a certain mutation" (Barthes, "Work" 155). As it turned out, the mutation that Barthes saw as gripping the idea of the work was held in check by the limited conceptual repertoire of the textual science in its structuralist phase. By contrast, in its new dispensation, textual science can assume its rightful place within the more general endeavor of cognitive science, under whose auspices a number of disciplines have begun to converge on the question of how people use language, in socially situated ways, to build, revise, and communicate models for understanding the world. Yet--and this is the crucial point--the mutated literary text still needs to be studied as a specific type of text. Barthes himself eventually opposed the concept of textuality to what he viewed as an outmoded and reactionary attempt to categorize certain kinds of texts as "literary" ("Work" 157-58). In shifting from a hermeneutics of the work to the sciences of the text, however, researchers do not ipso facto commit themselves to a denial of the specificity of literature. On the contrary, theorists can finally begin to come to terms with literature's unique properties. Those properties derive from the way literary discourse unfolds as a specific kind of situated textual practice, a form of practice that organizes how we think and act within a particular region of sociocommunicative space.[13] Department of English North Carolina State University dherman@unity.ncsu.edu ---------------------------------------------------------------- COPYRIGHT (c) 2001 DAVID HERMAN. READERS MAY USE PORTIONS OF THIS WORK IN ACCORDANCE WITH THE FAIR USE PROVISIONS OF U.S. COPYRIGHT LAW. IN ADDITION, SUBSCRIBERS AND MEMBERS OF SUBSCRIBED INSTITUTIONS MAY USE THE ENTIRE WORK FOR ANY INTERNAL NONCOMMERCIAL PURPOSE BUT, OTHER THAN ONE COPY SENT BY EMAIL, PRINT OR FAX TO ONE PERSON AT ANOTHER LOCATION FOR THAT INDIVIDUAL'S PERSONAL USE, DISTRIBUTION OF THIS ARTICLE OUTSIDE OF A SUBSCRIBED INSTITUTION WITHOUT EXPRESS WRITTEN PERMISSION FROM EITHER THE AUTHOR OR THE JOHNS HOPKINS UNIVERSITY PRESS IS EXPRESSLY FORBIDDEN. THIS ARTICLE AND OTHER CONTENTS OF THIS ISSUE ARE AVAILABLE FREE OF CHARGE UNTIL RELEASE OF THE NEXT ISSUE. A TEXT-ONLY ARCHIVE OF THE JOURNAL IS ALSO AVAILABLE FREE OF CHARGE. FOR FULL HYPERTEXT ACCESS TO BACK ISSUES, SEARCH UTILITIES, AND OTHER VALUABLE FEATURES, YOU OR YOUR INSTITUTION MAY SUBSCRIBE TO PROJECT MUSE, THE ON-LINE JOURNALS PROJECT OF THE JOHNS HOPKINS UNIVERSITY PRESS. ---------------------------------------------------------------- Notes 1. See Dosse for a comprehensive and highly readable account of the vicissitudes of the structuralist revolution in which Barthes participated. In "Structuralism's Fortunate Fall" I examine Dosse's history of structuralism in light of the broader problem of writing nonreductively about the history of literary and cultural theory. 2. An important exception in this regard is Thomas G. Pavel's The Feud of Language, which offers a critical reappraisal of the structuralists' appropriation of linguistic concepts and methods. 3. An earlier version of the second section of my essay appeared as "Dialogue in a Discourse Context." 4. See my "Roland Barthes's Postmodernist Turn" for a fuller account of the "postmodern turn" that led to Barthes's rejection of textual science as a research goal. 5. For an early argument to this effect, published only one year later than Barthes's essay on the structural analysis of narratives, see Hendricks's insightful account of discourse-level (i.e., suprasentential) properties of language. 6. On the anaphoric functions of definite descriptions, see Green (26-34). 7. For additional attempts to use pragmatic, discourse-analytic, and sociolinguistic models to analyze literary dialogue, see my "Mutt and Jute" and "Style-Shifting." 8. See J.J. Gumperz (100-29), Deborah Schiffrin (56-7), John Searle (30), and Deborah Tannen (18-19). 9. For a discussion of how Woolf's speech representations in Between the Acts similarly complicate the very idea of speech acts, see my Universal Grammar (139-81). 10. See Hymes's (51-62) original presentation of the SPEAKING grid, and, for an elaboration and refinement of Hymes's model, Muriel Saville-Troike's excellent textbook on the ethnography of communication. 11. Analogously, Schiffrin examines speaking for another as a particular sort of alignment strategy, i.e., a way for interlocutors to chip in rather than butt in (106-34). Schiffrin discusses the bearing of this alignment strategy on gender roles. 12. See my "Story Logic" for a preliminary attempt along these lines--one that outlines a basis for comparing and contrasting literary and conversational narratives. 13. I am grateful to James English and to an anonymous reviewer for comments and criticisms that helped me revise an earlier version of this essay. The hard questions put by the reviewer proved especially helpful as I tried to clarify my argument. Works Cited Austin, J. L. How to Do Things with Words. Oxford: Oxford UP, 1962. Barthes, Roland. "The Death of the Author." Image, Music, Text. Trans. Stephen Heath. New York: Hill and Wang, 1977. 142-48. ---. Elements of Semiology. Trans. Annette Lavers and Colin Smith. New York: Hill and Wang, 1968. ---. "From Work to Text." Image, Music, Text. Trans. Stephen Heath. New York: Hill and Wang, 1977. 155-64. ---. "Introduction to the Structural Analysis of Narratives." Image, Music, Text. Trans. Stephen Heath. New York: Hill and Wang. 79-124. ---. "The Structuralist Activity." Critical Theory Since Plato. Ed. Hazard Adams. San Diego: Harcourt Brace Jovanovich, 1971. 1196-99. Brown, Penelope, and Stephen C. Levinson. Politeness: Some Universals in Language Usage. Cambridge: Cambridge UP, 1987. Chomsky, Noam. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press, 1965. ---. Syntactic Structures. The Hague: Mouton, 1957. Cohn, Dorrit. Transparent Minds: Narrative Modes for Presenting Consciousness. Princeton: Princeton UP, 1978. Derrida, Jacques. Of Grammatology. Trans. Gayatri Chakravorty Spivak. Baltimore: Johns Hopkins UP, 1976. ---. "Structure, Sign and Play in the Discourse of the Human Sciences." Critical Theory Since 1965. Eds. Hazard Adams and Leroy Searle. Tallahassee: U Presses of Florida, 1986. 83-94. Dosse, François. History of Structrualism. Vols. 1 and 2. Trans. Deborah Glassman. Minneapolis: U of Minnesota P, 1997. Goffman, Erving. Forms of Talk. Philadelphia: U of Pennsylvania P, 1981. Green, Georgia A. Pragmatics and Natural Language Understanding. Hillsdale: Erlbaum, 1989. Gumperz, J. J. Discourse Strategies. Cambridge: Cambridge UP, 1982. Hendricks, William O. "On the Notion 'Beyond the Sentence.'" Linguistics 37 (1967): 12-51. Herman, David. "Dialogue in a Discourse Context: Discourse-Analytic Models and Woolf's To the Lighthouse." Virginia Woolf Miscellany 52 (1998): 3-4. ---. "The Mutt and Jute Dialogue in Joyce's Finnegans Wake: Some Gricean Perspectives." Style 28.2 (1994): 219-241. ---. "Roland Barthes's Postmodernist Turn." Postmodernism: The Key Figures. Eds. Joseph Natoli and Hans Bertens. Oxford: Basil Blackwell (forthcoming). ---. "Story Logic in Conversational and Literary Narratives." Narrative (forthcoming). ---. "Structuralism's Fortunate Fall." Postmodern Culture 8.1 (1997); and . ---. "Style-Shifting in Edith Wharton's The House of Mirth." Language and Literature 10.1 (2001): 61-77. ---. Universal Grammar and Narrative Form. Durham: Duke UP, 1995. Hymes, Dell. Foundations in Sociolinguistics. Philadelphia: U of Pennsylvania P, 1974. Kant, Immanuel. Critique of Pure Reason. Trans. Norman Kemp Smith. New York: St. Martin's, 1965. Labov, William. Sociolinguistic Patterns. Philadelphia: U of Pennsylvania P, 1972. Levinson, Stephen C. Pragmatics. Cambridge: Cambridge UP, 1983. Lévi-Strauss, Claude. Structural Anthropology. Trans. Claire Jacobson and Brooke G. Schoepf. New York: Basic Books, 1963. Pavel, Thomas G. The Feud of Language: A History of Structuralist Thought. Trans. Linda Jordan and Thomas G. Pavel. Oxford: Basil Blackwell, 1989. Pinker, Steven. The Language Instinct: How the Mind Creates Language. New York: Harper-Collins, 1994. de Saussure, Ferdinand. Course in General Linguistics. Ed. Charles Bally, Albert Sechehaye, and Albert Riedlinger. Trans. Wade Baskin. New York: McGraw-Hill, 1966. Saville-Troike, Muriel. The Ethnography of Communication: An Introduction. 2nd edition. Cambridge, MA: Blackwell, 1989. Schiffrin, Deborah. Approaches to Discourse. Oxford: Blackwell, 1994. Searle, John. Speech Acts. Cambridge: Cambridge UP, 1969. Tannen, Deborah. Talking Voices: Repetition, Dialogue, and Imagery in Conversational Involvement. Cambridge: Cambridge UP, 1989. Woolf, Virginia. To the Lighthouse. San Diego: Harcourt Brace Jovanovich, 1989. Woolgar, Steve. "Reflexivity is the Ethnographer of the Text." Knowledge and Reflexivity: New Frontiers in the Sociology of Science. Ed. Steve Woolgar. London: Sage, 1988. 14-34. ---. Science: The Very Idea. New York: Tavistock, 1988.