Linguistics and Literary Study

Journal of Advanced Composition 13/2, 1993, 423-448.

Closing the gap between linguistics and literary study:

Discourse analysis and literary theory[1]

Robert de Beaugrande

It is a commonplace among faculties and departments of language that linguistics and literary studies are distinctive domains, and that their interaction tends to be limited. But it would be unjustified to attribute this divergence to mere feelings of mutual rivalry, insecurity, or mistrust. Instead, the two domains have differed so fundamentally in their traditional conceptions and directions that immediate interaction has been difficult on purely logistic grounds. I shall undertake to indicate first why this was so and then why recent fundamental changes in both domains give reason to believe that conditions are now much more auspicious for concerted interaction.

The relation or tension between the ‘paradigms’ whereby ‘business as usual’ was for a long time conducted in linguistics and literary studies was unlikely to be assessed as long as the paradigms remained implicit and worked fairly smoothly. Both fields typically proceeded on the assumption that they could get on with their business and not to disclose and legitimise what they were up to. This tactic presupposes that the participants share a fairly firm and constricted consensus about what should be done and how to go about it. Such a consensus may be productive for a time, but, if the object of inquiry is as complex and diverse as either language or literature, sooner or later grows uncomfortable and leads to stagnation.

In addition, the ‘participants’ to be considered here include not just the professionals, the linguists and literary scholars, but also the ‘clients’ who turn to us for expertise to apply to the learning and teaching of language or literature. Since the onset of the 1960s, we have witnessed such a profound transformation of our ‘clientele’ that we have been compelled to lay aside our reassuring certainties and scan new horizons. In this frame of mind, we soon began to see that when our underlying ‘paradigms’ were explicitly displayed, they did not fit the needs of the new clientele nearly as well as we would like to think. This changed situation is the common impetus for recent trends, even when the latter seem quite diverse or oriented toward more abstruse or academic concerns.

My brief will accordingly be that a reassessment of the prospects for interaction is needed not merely because the two fields themselves have made impressive progress in theory and research, but also because we urgently need a framework to design integrated language programs for a rapidly evolving ambience. To a certain degree, which I shall attempt to clarify, the convergence has not been deliberate or co-ordinated. But I feel it both can and should be in the coming years, when the challenges on both fronts are virtually certain to become more complex and diversified.

Traditional contrasts

To appreciate why this was so, it might be helpful to contemplate a schematic set of contrasts such as that presented in Table 1.

Like most heuristics, this Table simplifies issues and irons out variety. Nor are there precise criteria for determining what time span these ‘traditions’ cover, but a very rough approximation might be 1880-1970.[2]

Whereas the object of linguistics was the language (Saussure’s ‘langue’) as an abstract system, the object of literary studies was the literary text as a concrete artefact. Admittedly, this explicit contrast entails implicit contacts: the language can only be inferred from a corpus of texts, among which literary ones are often influential, while the literary text must be an instantiation of its language. But these contacts remained largely submerged or taken for granted, and, as we shall see, they can be quite problematic.

For the linguist, material was to be derived from data, which included all the samples of the language that could expediently be assembled and collated by means of fieldwork and later, for the generativists (whose stance toward fieldwork remained uneasy), by means of introspection. For either source, all the data -- aside from special cases (e.g. modernist poetry) or errors in transcription -- were held to belong equally to the language. For the literary scholar, material came from the ‘canon’ of literary texts established mainly by tradition, witness the long-standing practice of anthologising meritorious (‘great’) works for public edification. The canon periodically underwent quiet revisions as some author or work was admitted or excluded, but the legitimacy of having a canon and the prerogative of literary studies to establish and cultivate it was not seriously challenged.

To stress its shift away from historical ‘philology’, modern linguistics programmatically adopted a synchronic approach by viewing the language as a system in its current state rather than in its evolution. Linguists like Saussure conceded that this ‘static’ construct was a fiction, since language is always changing; but they saw no other way to design theories and models that fit their sparse notion of ‘system’. Literary studies, in contrast, remained resolutely historical, witness such time-honoured conventions as organising the program or personnel of literature departments by century or period and treating contemporary literature at best marginally alongside the canon of ‘classics’ of the past.

Linguistics worked with a construct of the ideal speaker who ‘knows’ the language and can produce an unlimited set of utterances (or ‘sentences’). The term ‘ideal’ was made fashionable by the generative paradigm (e.g. Chomsky), but the ‘speaker’ envisioned at least since Saussure had unmistakably been an idealisation. Literary studies, in contrast, was concerned with the real author as a biographical and historical figure, and intense effort was expended on documentation, e.g. through official records, personal letters and diaries, contemporary comments, dates and places of publication, and so on.

At the other end of the transaction, linguistics assumed an ideal hearer (often just called ‘hearer’) who possesses essentially the same knowledge as the ideal speaker and can understand the same set of utterances. When intuition came into vogue, the linguist was entitled to stand in for both speaker and hearer, inventing sample sentences and rendering interpretations -- in retrospect, a step backwards. In literary studies, however, we notice from the beginning a significant vacancy: the role of the reader was usually not addressed as an issue but tacitly occupied by the scholar, whether an academic or a professional critic, who purported, by virtue of status, to be the proper (qualified, discerning, etc.) reader for the literary work. Intriguingly, the traditional move was to present one’s own reading in the name of the real author, e.g. of what Shakespeare or Milton ‘was saying’, ‘meant’, ‘intended’, etc., and thus to merge author with authority, if not indeed with an authoritarian posture. It thus seemed unnecessary and distractive to treat one’s own reading as just one instance among many others, or as a statement of individual or personal response.

Whereas linguistics was concerned with the entire community of speakers, which the generative paradigm expressly declared to be ‘homogeneous’, the widest group addressed in literary studies was usually the school or movement to which an identifiable set of real authors could be assigned by conspicuous stylistic or thematic attributes, e.g. ‘Barock’ in Germany and ‘Gongorrismo’ in Spain. The name was often a label attributed in hindsight by the scholars rather than designation devised by the authors themselves. But either way, a name once bestowed tended to become an integral category of literary studies, especially for historical and pedagogical purposes.

As signalled by Saussure’s landmark title Cours de linguistique générale, linguistics sought to formulate the most general principles, for which the ‘laws’ of the ‘sound shifts’ formulated by philology had provided the most shining examples. A premium was placed on generalisations applying to an entire language, or, better still, to all languages (‘universals’). In literary studies, much attention was accorded to the special or even unique quality of the literary work, and the high regard for detail could be seen in the common exercise or test for students of memorising or identifying individual poems or passages from plays, novels, etc.

Linguistics addressed the rules of language encoding the patterns, usually formal, which apply to all or most instances, e.g. the placement of ‘article’ before ‘noun’ in English. Literary studies addressed the conventions of genre, some of them based on form (e.g. for the ‘sestina’) and some based on theme or topic (e.g. ‘revenge tragedy’). Certain trends, such as Russian Formalism and American New Criticism, have sought to bridge this contrast by detailed formal analysis of certain genres, but the results have remained disputatious, largely because of the problematic implication that ‘literariness’ or ‘poeticity’ is something ‘in the language’ of the text (cf. third section).

To the degree that it was influenced by linguistics, the study of style centred on the notion of choice, i.e. the selection of certain options offered by the overall language system. In the literary domain, the notion of style as ornamentation persisted, i.e., an aesthetically pleasing addition of ‘schemes’ and ‘tropes’ which students should be taught to name and identify with erudite terms like ‘synecdoche’. The ‘content’ or the ‘message’ of the work was typically held to exist apart from this ornamentation.

In programmatic opposition to traditional grammars, linguistics resolved to be non-evaluative, recoding and describing language irrespective of prescriptive and proscriptive attitudes about ‘good’ and ‘bad’ or ‘correct’ and ‘incorrect’. Literary studies has remained evaluative, despite occasional declarations that values tend to obscure or distort; after all, the mere choice of a text for analysis and interpretation already implicates a value judgement.

The goal of linguistics was the description of a whole language as a total system, a characterisation of its phonological, morphological, and grammatical regularities in a compact and perspicuous format. The generativist arguments ranking ‘explanation’ over ‘description’, though an intensely polemical issue for a time, did little to change this goal, witness the specialised definition of ‘generating a sentence’ as ‘assigning to it a structural description’.[3] What was offered up as ‘explanation’ usually turned out to be a structural description, and a non-committal one at that. The goal of literary studies was to a large extent the advocacy of one’s interpretation of a particular work and, in conjunction, of the work itself as a meritorious exemplar worthy of such explication.

Statements and claims in linguistics were confirmed by data as additional samples were collected and compared to a given formulation. Since potential data are infinite, it could not be determined exactly how much data were need or confirm or disconfirm; and linguistics has been replete with formulations that were later found to be premature. In literary studies, the implicit standard for confirming an interpretation was the eloquence of the scholar in persuading, convincing, and creating harmony and order.

Prospective linguists underwent training by method, the most noteworthy being the techniques for eliciting, recording, and analysing data by fieldwork. The success of descriptive and ‘tagmemic’ method in constructing grammars for remote languages, sometimes even without the aid of bilingual informants, surely constitutes the most enduring and admirable achievement of the discipline. Prospective literary scholars were traditionally trained by imitating the interpretive performances of established scholars, including their teachers, upon concrete works. Whereas linguistics was characterised by collective research among teams and each contributor sought to expand or stipulate the accumulating model (or ‘grammar’), literary studies was devoted to individual research, and each contributor sought to overturn previous interpretations of the same work.

Linguistics has had a reputation for being theory-centred, though this is not fully justified in view of the enormous practical achievements in descriptive fieldwork. Still, the theoretical aspects have been widely emphasised, partly in tribute to scientific decorum and partly to dissociate the linguist from the amateur or ordinary student of language. Literary studies, on the other hand, has had a reputation for being practice-centred, based firmly on the activities of reading and interpreting rather than on the formulation of abstract principles, though this too is not fully justified in view of the steady input from philosophy, aesthetics, history of ideas, and so on.

These, then, are some traditional contrasts between linguistics and literary studies which help to account for their lack of direct interaction in past decades. The occasion to rehearse them is the major shifts which, during the last twenty years or so, have profoundly unsettled the conventions on both sides in ways that create an auspicious scenario for a fundamental reconciliation. I would argue that these shifts have resulted more or less spontaneously from the increasing pressure of unresolved problems generated on both sides by the standing conventions I have outlined. Indeed, one might go so far as to argue that, if pursued without regard for the consequences, these conventions could lead to untenable positions and to a crisis between theory and practice. But scholars on both sides have admitted various compromises or modifications, albeit more often been implicit than programmatically declared.

Text and Discourse as ‘Linguistic’ Entities

The celebrated distinction between language and discourse, or ‘langue’ and ‘parole’, or system and usage, is one the most fundamental conceptions of modern linguistics, but also one of the most disputatious. Though he probably intended to protect linguistics from absorption by neighbouring sciences, the resolution at the end of Saussure’s Cours -- ‘the true and unique object of linguistics is language studied in and for itself’ -- encouraged scholars to believe that the two sides could and should be kept separate. Yet as an empirical human phenomenon, the language itself is never given or present for observation, nor is it known to any one speaker. What is given and present is always the text or discourse, and whatever any one speaker ‘knows’ of the language must be an abstraction and summation from experiences with text and discourse. Even the isolated sentence treated as an object for analysis is part of the discourse of the analysis; its context is not dissolved, but merely transformed.

To use the term ‘language’ in the theoretical sense established by Saussure is to appeal to a hypothetical consensus among the community of speakers despite their inevitable diversities of knowledge and experience. In effect, linguistics instated, without empirical justification, the credo that this consensus forms a de facto adequate basis for general statements and does not constitute a serious theoretical problem in its own right. This credo remained fairly intact as long as the main emphasis fell on those issues in phonology, morphology, and grammar that constitute focal points of regularity, but became unsettled as research progressed toward less regular issues and devoted more attention to the domains of semantics and pragmatics.

A particularly forceful thrust into new waters came when ‘grammar’ was reinterpreted by the generativists in the much more ambitious sense of a complete set of structures and rules which describe all sentences of a language and exclude all non-sentences. The inaugural hope that a limited domain of pure ‘syntax’ or grammar could be fenced off and described independently proved elusive as soon as research progressed beyond the introspective cases hand-picked to fit the approach. The attempts to formulate such a grammar encountered substantial diversity where consensus had long been assumed. It became clear that syntax alone could not supply the needed constraints, and semantics and pragmatics would have to be integrated. Moreover, as long as the generativists did not have a conception of text or discourse, they faced the daunting if not impossible task of attempting to state, at the level of the virtual system, all the constraints that could apply to actual utterances; or, failing that, of determining, by precise, motivated, and practicable criteria, just which constraints were or were not relevant.

The turn toward text linguistics and discourse analysis in the 1970s and 1980s was thus to some degree less an attempt to break genuinely new ground than a response to pressure from problems inherent in ‘non-textual’ (sentence’) linguistics. Hence, early text linguists imagined that their task would be to add a complementary set of rules and structures onto the ‘sentence grammars’ proposed so far, so as to account for the linkage of sentences into sequences, the distribution of pronouns, the formation of extended ‘referential chains’ and so on. This project only made the task that much harder, and the resulting ‘text grammars’ were yet more complicated and ambitious; the task was still to state, at the level of the ‘text grammar’ (abstract system), all the constraints that could apply to texts.

The insight only gradually emerged that working at this level of abstraction was self-defeating for text linguistics. The generalisations that can be made about all texts are not terribly rich or enlightening, and the focal points of regularity in phonology, morphology and grammar constitute unmanageable degrees of detail for text analysis. To attain more powerful and unifying methods, we would have to address types of texts and conditions of text production and reception. We confront so much data in text and discourse that exhaustive treatment, either in the minimal units of ‘structuralism’ or the formal structures of ‘generativism’, was neither readily feasible nor particularly informative. Instead, we would have to proceed from the focal points of control, such as topic, goal, and situational context, in order to determine which units or structures are the more relevant ones for a given concrete domain.

In effect, the backlog of problems fomented by the original Saussurian distinction finally had to be faced:[4] linguistics would have to provide not just theories and models of language (‘langue’, ‘grammar’, etc.), but theories and models which could show how knowledge of the language, including that gathered and ‘systematised’ by linguistics itself, can emerge from experience with text and discourse; and, conversely, how text and discourse are ‘actualised’ in respect to the language as well as to other relevant cultural, social, and psychological factors. That linguists might be reluctant to embark on such an enterprise is readily understandable; but I see no other prospect for material progress which could free the discipline from the stagnation and fragmentation we have witnessed in so many areas since 1970.

In respect to the standing conventions outlined in Table 1 for conventional linguistics, the turn to text and discourse entails a range of shifts, as characterised in Table 2 . Though they emerged from different ambiences, ‘text linguistics’ and ‘discourse analysis’ have converged today to the extent that we can treat them together for the purposes of a programmatic survey.[5]

Instead of the entire language as an abstract virtual system (repertory of potential choices) envisioned by general linguistics, text linguistics takes the text and discourse as the basic entities, where ‘discourse’ straightforwardly designates a set of texts directed to each other, especially in conversation. The text is neither an abstract system nor a concrete artefact, but an interactive event which itself has the character of an operating actual system (array of choices actually made). The current function of a given element in the text-system is determined partly by its function in the abstract system and partly by the current functions of co-occurring elements in that context. Thus, the potential meaning of a lexical item in the lexicon of English acts a set of ‘parameters’ which are specified or adjusted when the item is assigned a meaning within a text-event.[6]

Whereas general linguistics uniformly seeks data of the entire language, text linguistics seeks data provided by particular text types and data regarding the production and reception of texts. As noted in above, this shift was enforced by the embarrassment of riches that texts provide and by the sparsity of statements that can be made about all texts. To be sure, the shift brings us fresh problems in formulating workable typologies of texts and realistic models of text processing, but impressive headway has been made in numerous areas.[7]

An emphatic turn away from introspection as the main source of data not merely reinstated fieldwork, but supplemented it with three further methods: participation, where the investigator joins in the practices of discourse as a social and cognitive agent, albeit one with special focuses and motives; experiment, where the investigator designs controlled discursive tasks, such as retelling a story in one’s own words; and simulation, where the investigator builds a working model to run on a computer, e.g. a story-reading program that can answer questions or make summaries.

The essentially static synchronic perspective held in place since Saussure is supplanted by a perspective which is not simply ‘diachronic’ (centred on the history and change of the whole language) but dynamic and procedural, centred on the ongoing discursive practices as they unroll in social interaction. The text loses its apparent obviousness as a written artefact and is posed as a problem of how it could be produced and received with relative ease and success despite the undeniable complexities involved. We must assume, for example, that memory storage is not unlimited, but efficiently organised to access and activate the materials needed for the ongoing procedures. Also, the ‘function’ of a text element (in the sense explained above) must fluctuate according to the stage of the discourse where it occurs and the contextual factors relevant at that stage.

The ideal speaker and the ideal hearer shift to the text producer and the text receiver as social and cognitive agents, i.e. as ‘whole human beings’ within a cultural setting who engage in discourse interaction in order to pursue goals and to gain or provide access to knowledge. These notions are still abstractions but are far more proximate to real text producers and receivers than were the ideal speaker and hearer, who were held to ‘know the language perfectly’ and to have no ‘memory limitations’.[8]

In place of the community of all speakers, which general linguistics had assumed to be uniform and homogeneous, we contemplate the community as a social complex of diverse classes and groups, among whom power and solidarity are unevenly distributed within the ‘prevailing order’. Their respective interests are typically asserted or denied by means of discourse, so that we can expect to find conflicts and contradictions where general linguistics tended to see a harmonious abstract system in which ‘everything is held in place’ (‘un système oú tout se tient’).

The general outlook that sought rules and regularities of the widest possible scope is now being reshaped as a cautiously monitored balance between general and specific. Here, we do not assume too readily that our data represent the whole language, but attempt to determine, by empirical means, how far it may be specific to a text type, social group, situational setting, and so on. Nor are specific data considered less valuable, informative, or ‘scientific’, since they materially help us bridge the gulf between the single text and the whole language.

The rule of linguistics had increasingly come to be seen as a formal algorithm for creating, describing or transforming patterns in sentences, much as a mathematical operation or a computer program might do. The strategy of text linguistics, in contrast, is a procedural heuristic for managing topics and goals in situations that, at some level of detail, are always novel. Whereas an algorithm is mechanical and guaranteed to yield the ‘correct’ result but applies only within strict limits, a strategy does not always work but is flexible and powerful enough to handle many contexts and needs.[9]

The notion of style as choice in linguistic stylistics is enriched by the notion of style as a mode of discursivity with concrete social consequences. The strategic use of one style over another offers an important means for pursuing goals and providing -- or denying -- access to knowledge (cf. last section). Clearly, style can no longer be treated as a matter of language alone, but a relation between language options and their characteristic motivations and effects.

The non-evaluative stance linguistics had adopted to dissociate itself from the prescriptive and proscriptive stance of traditional grammars is revised to be evaluative, but by interactional criteria rather than vague attitudes about ‘good’ and ‘bad’ or ‘correct’ and ‘incorrect’. These criteria must be demonstrably relevant to the success of communicative events. A textual usage counts as efficient if it is easy to handle, effective if it helps toward achieving a goal, and appropriate if it suits the occasion. ‘Good usage’ and ‘correct grammar’ may not qualify by such criteria, e.g. if they encourage complicated syntax or flowery diction that only makes the audience confused and irritable.

If the goal of linguistics has usually been description (even when it was offered as ‘explanation’) (p. 2f), the text linguistics and discourse analysis have the further goal of application to discursive practices. Our highest priority would be to enhance the freedom of access to knowledge through discourse, thereby aiding people to grasp both the world they live in and their opportunities to develop themselves in education, career, and personality. ‘Knowledge of the language’ in the sense of general linguistics is obviously just one factor, and if it were indeed fully uniform or ‘homogeneous’ it could not the crucial one. However great the consensus about the whole language, discourse strategies for expressing knowledge, obtaining co-operation, making a favourable impression, and so on, are special skills quite unequally distributed, especially for commanding a range of styles.

The convention of confirming generalisations, rules, and so on by confronting them with data is of course still paramount in text linguistics, but an even more crucial test is whether our findings are confirmed by social relevance. If we are working to enhance the freedom of access to knowledge, then we must obtain concrete results in such areas as education, training, terminology, and translation. In this sense, we must take our own advice and strive to make knowledge about text and discourse accessible to those who require it, e.g. to designers of educational materials. A higher concern for the readability and accessibility of our own discourse about discourse may counterbalance the somewhat forbidding and abstruse quality of some linguistic treatises in the past.

Similarly, the training by method put in place by descriptive linguistics remains an integral part of our enterprise, though not in the stance of the ‘detached scientist’ invoked, say, by Bloomfield or Z.S. Harris. Instead, training must explicitly include the prospective investigator's engagement with the data in its social context, and thus address the motivation for doing one type of research rather than another. The detached stance of traditional science claimed to be neutral and non-political, but in effect worked in favour of the status quo and the institutional powers that sustained it. Today, text linguistics and discourse analysis are increasingly engaged in a ‘critical’ function of seeking and providing the knowledge that make it possible to change the status quo and resist manipulations by institutional powers not merely in government and administration, but in commerce, mass media, and so forth.[10]

Collective research is at least as necessary in text linguistics as in general linguistics, if not more so in view of the expanding interdisciplinary scope. However, this research needs to be carried out as a self-reflective activity continually contemplating its own conditions, including those of producing specialised discourse about general discourse.[11]

Finally, the theory-centred reputation linguistics had taken on, especially during the ascendancy of generativism, is now yielding to a monitored balance of theory with practice. Like that between general and specific, this balance is enforced by the sheer necessity of designing any theory at all, since, without reference to practice, the set of possible theories of text and discourse is unmanageably broad. Thus, a theory of text as a formal array of the ‘deep structures’ of its sentences would not merely be explosively complicated but only distantly related, say, to a model of readability for schoolbooks. A theory of text in terms of cognitive processing, in contrast, is relatively proximate and has already been shown to have practical relevance in this area. [12]

The shifts summarised in Table 2 and briefly outlined in this section are therefore indicative not of some coincidence of trendy fads, but of an integral evolution required for significant progress by any standards, including those to which linguistics itself is fundamentally committed. To complain that the newer theories and models do not meet predetermined criteria of abstractness and formality is to perpetuate the folk wisdom of ‘science in a vacuum’ that draws no conclusions when science makes possible the voracious technological depredation of the planet and the imminence of global nuclear destruction. We will need a reputable and comprehensive body of data, issues, and projects before we can state what criteria are in fact appropriate to the imposing tasks ahead.

Literary Theory

‘Literary theory’ has become a cover-term for an increasingly diffuse trend away from a concern with the individual text toward a concern with the general conditions of literature or ‘literariness’. To some degree, this trend might appear to the complement of that described for the shift toward text and discourse in linguistics, since generality is being lowered in the first case and raised in the second. Moreover, we find some commonalities in the emerging terminologies on the linguistic and literary side: not just ‘text’ and ‘discourse’ themselves but ‘discourse analysis’, ‘textuality’, ‘intertextuality’, and so on. However, the uses of these terms diverge as do the contexts in which they appear. Indeed, one can find two contemporaneous introductions to ‘Discourse Analysis’ with virtually no sources in common.[13]

And in terms of evolution, the parallels are again imprecise. Literary theory resolved to strip away the obviousness and routine of literary studies and its ‘object’, and to turn whatever came to light ‘underneath’ into a theoretical problem for new inquiry. This sensitivity toward submerged problems at the very base of the discipline came earlier and developed more scope and momentum in literary studies than in linguistics. The major motive for the difference was surely the complexity the literary text presents to professional inquiry, as compared to the simple isolated sentences so popular in linguistic discussions e.g. ‘the farmer killed the duckling’ (Sapir) or ‘John is easy to please’ (Chomsky). Literary scholars could easily see from the enormous acuity expended on any ‘great work’ that the issues involved were complex, though not in the sense of linguistics, e.g. as an agglomeration of minimal units or a derivation of multiple sentences from ‘deep structure.

The Saussurian resolution that ‘the true and unique object of linguistics is language studied in and for itself' had no counterpart in literary studies. To proclaim that its ‘true and unique object is literature studied in and for itself’ would have seemed either gratuitous (if one is denying the notion, which nobody has seriously affirmed, that literature should be studied merely as source material for history, theology and so on), or else arrogant (if one is ordaining that literature should be cut off, a priori, from the history of culture and ideas). Moreover, it would be premature to imply that we in fact have a well-defined notion of ‘literature in and for itself’. On the contrary, the attempt to draw borders around it seems singularly unproductive; it is hard to imagine a single human concern or theme which has not been evoked by literature at some time.

The early thrust toward literary theory, as exemplified by Formalism, New Criticism, and such schematics as Wellek and Warren’s Theory or Frye's Anatomy, were motivated chiefly by a desire for an explicit and organised methodology, whether the inspiration came from linguistics (as with Formalism) or mythology (as with Frye). Whereas intuition was increasingly enshrined in linguistics by the generativists, it was increasingly questioned in literary studies, albeit much more gradually. The institutions of literature and its study were after all vastly larger and more entrenched than linguistics even at its highest points; and literary theory did not enjoy the added advantage for the generativist school, which arrived on the scene just when many new linguistics programs and departments were opening and got in on the ground floor, as it were.

But the truly decisive momentum of literary theory that changed it from an abstruse specialisation within a few ‘comparative literature’ programs into an internationally prominent topic in a great majority of language and literature programs came when the search for methodology turned into a comprehensive engagement with the problematics of literature as a whole, especially with those that were making the search so arduous and elusive. The accessibility of the literary text as a written artefact ‘on the page’ -- a view which Formalism and New Criticism had if anything reinforced -- was increasingly put in question, and an unwonted participant in the literary transaction took centre stage: the reader, whose role had for centuries been tacitly occupied by any scholar or critic who wished to assume it. This reader became the focal territory of all the complexities and perplexities of literature in an engagement with the text that displaced ‘the text itself’ as the focus of attention and the phenomenon to be accounted for.

Probably because their first interest in literature had been mainly historical, most of the earlier champions of the ‘reader’, such as Jauss, Iser, and Fish, retained the author as a concrete personage and point of authority for arguing that one way of reading was more suitable than another -- that the issues uncovered were intended and designed by Baudelaire, Fielding, or Milton. Even where other frameworks of authority were offered, such as Freudian psychoanalysis (e.g. Holland), Marxist theory (e.g. Jameson) and feminism (e.g. Millett), the real author characteristically remained in place, though less in control than implied by the historical groundings, e.g. as a locus of psychic drives, class conflicts, or sexual politicking.

Once the act of reading was acknowledged to portend complex problematics, the act of writing was bound to be reconsidered along similar lines. In the mid-1960s, E.D. Hirsch could still argue that writing was ‘determinate’ but reading was ‘whimsical’ and ‘lawless’, without sensing the incoherence of his position nor the potential of his methods of ‘validation’ for burying the literariness of literature beneath a naive and ponderous scholarly apparatus whose ‘scientific’ credentials rested on the misguided notion, borrowed from philosophy of science (e.g. Popper), that one can test and validate hypotheses without inquiring where they come from.[14]

But by the mid-1970’s, the model of a precisely circumscribed author seemed incongruous vis-à-vis the influential models of a perplexed and self-doubting reader, especially when we contemplate such authors as Rousseau or the English Romantics. Hence, literary theory finally began to turn from the real author toward models of the author which any real author would fit only approximately. Here, authorship is more a performance and a goal than a state or attribute of a person.

No longer shackled to historical biography, this model of the author could only come from engagements with literary texts. If, as Iser effusively showed, the text entails an ‘implicit reader’[15], then it can equally well entail an ‘implicit author’ who is just as much a literary conception as is a fictional character in a play or novel. The heavy investment of traditional studies in biographical documentation tends to dull our awareness of the degree to which our notion of all literary authors flows first and foremost from their work, whether they conspicuously centred their opus on creating a vision of themselves (e.g. Whitman) or whether they were cryptically reticent about themselves (e.g. Shakespeare).

The overall progression theorising thus relentlessly led to a programmatic destabilising of the classic triad of text, reader, and author up to the point where uncertainty, ambiguity, figurality, and the like were no longer obstacles for the scholar to resolve, but essential factors to be enacted or even celebrated. Startling disruptions of academic decorum became fashionable, though chiefly by authorities who, like Bloom and Hartman, could afford them by virtue of their prior careers as traditional scholars.[16]

Yet in a certain sense the wheel had merely come full circle. The long-standing notion of the literary text as an ‘object made out of a language’ had been problematic from the start, but had supported an expansive enterprise purporting to analyse, interpret, and explain such ‘objects’ in emulation of philosophical, historical, and scientific methods. The obvious fact that this enterprise did not seem to be producing definitive results was either ignored or else explained as a temporary inconvenience we could eliminate when we had gathered enough examples and perfected our methods -- a belief still underlying Hirsch’s project of validation which, thankfully, is not on the agenda (not even on his).

In contrast, the insight that literature is not a set of such ‘objects’ but a mode of discursivity and engagement, though far more appropriate and productive, could not have seemed auspicious as long as ‘discourse’ itself was not a prime theoretical entity either in literary studies or in philosophy, history, and science. The expanding preoccupation with discourse in various guises -- within ‘ordinary language’ philosophy, historiography, ethnography, communication, conversational analysis, social psychology, cognitive science, psychotherapy, pedagogy, and many more areas -- create an auspicious ambience in which literary studies could be reconciled with what it always had been but had often felt uncomfortable about being.

Today, there is nothing particularly outlandish in asserting that ‘literature’ is a communicative domain for creating and contemplating alternative worlds; even a realistic or documentary reconstruction of reality shows us the ‘real’ as one among a set of alternatives. Or in asserting that poetry extends this principle to language itself by practising alternative uses, or by displaying ordinary uses as one alternative. Moreover, the recognition that the aesthetic aspect of this engagement arises from the persistence of multiple interconnected significances means that interpretation cannot be the imposition of harmony and order alone, but the enactment of a dialectic between harmony and conflict, between order and incongruity, between real and potential -- which brings us to much the same standpoint as discourse analysis. This dialectic refers to and subsumes the diversity of elements without negating it, so that the language material deployed for a given ‘style’ cannot subsist independently from the ‘content’ or ‘message’ yet is not identical with it either as means or as effect. Similarly, values do not obscure or distort but constitute both precondition and ambience of the engagement.

If it follows that literary theory should recognise and account for this ‘alternativity’ and dialectic in its models of text, reader, and author, then the development I have sketched seems reasonable, if not indeed compelling. But we will need a different epistemology which can only emerge from a higher-level consolidation of the models in literary theory (and in discourse analysis as well). Close scrutiny reveals a disquieting number of literary theorists -- the examples are too glaring to need being named here -- for whom theory is just one more means for personal aggrandisation, if not for an outright personality cult; and the old battle over who has the ‘right’ interpretation threatens to be succeeded by an equally acrimonious battle over who has the ‘right’ model of author, text, or (especially) reader.

In sum, we can indicate the shift from traditional literary studies toward literary theory with the parameters shown in Table 3.

Admittedly, the simplifications involved here are especially acute, because literary theorists place a much higher value on individualism than do text linguists.

The text as a written (and presumably closed) artefact is ‘decentred’ into discourse as an open-ended transaction, which for some theorists (e.g. Foucault) extends to broad social and institutional frameworks. The term ‘intertextuality’ has gained some currency for the vision of the ‘open’ text as a meeting point or ‘weaving’ of other texts. Such a vision was traditionally either eschewed as detrimental to the reputation of a given work and author or else relegated to ‘influence studies’ of a somewhat antiquarian cast.

The canon as the established and accepted catalogue of works becomes the issue of how and why certain works or authors were selected, and how and why revisions of the canon came about. The contours of the canon were considerably relaxed, and new attention was accorded to ‘trivial’ and ‘popular’ literature as part of groundwork on which ‘high’ literature rested. A recent intriguing revision has been the authors who rose in fashion in the wake of literary theory itself because, like Rousseau or Shelley, they so aptly illustrated the perplexities foreseen by such models as those of the now-scattered ‘Yale school’.[17]

The notion of literature simply being handed down by tradition is displaced by an examination of the ways it is mediated and channelled by literary institutions, including not just the ‘academy’ of literary studies in universities and institutes, but the policies of publishers and editors, the awarding of literary prizes, and so on.[18]

The historical orientation, which projected a view of literature in a chronological progression of authors and works, shifts toward a programmatic orientation which sees literature as a complex of projects for navigating the complexities of literary communication. The orderliness of chronological methods is found to be a liability in disguising trends and currents across diverse time periods, e.g. the one linking German ‘Expressionismus’ of the 20th century with the ‘Barock’ of the 17th. Even the vision of an author influencing another who came earlier (e.g. Shelley’s Cenci as a tribute to Browning) has had a certain vogue: though historically perverse, it help us to perceive a richer orchestration of voices among alternative means toward similar ends.

For the motives sketched above, the real author was gradually, and by no means unanimously, recast as the more or less general model of the author. A corollary was the emergence of literary production as a category for describing what functions such a model might assume, and in some theorising (e.g. Foucault’s), came to dominate over the author, who was progressively ‘de-centred’. This occurred, I have suggested, in symmetry with the prior shift whereby the scholar ceased to be automatically instated as reader and a model of the reader was propounded. Here, we witness a corresponding turn toward the category of literary reception, or, in Germany, ‘aesthetics of reception’.[19]

If the school or movement had been a traditional means of categorising, the more complex notion of ‘horizon’ (to borrow an influential term propagated by Jauss) subsumes all the factors bearing on what authors and works were expected to involve. Whereas ‘school’ or ‘movement’ suggests an often misleading unity, ‘horizon’ suggest the background or frame, with the work in the foreground partly fulfilling and partly revising it. This dynamic conception altered the traditional concentration on the special or unique qualities of the single work in favour of a dialectic between innovation and expectation. The specific achievement was thereby seen not as some miraculous ‘aborigination’ or divine inspiration, but a strategic and highly skilled modification of prior systems of shaping and sense-making.

If the conventions of genre had been stressed for classificatory or pedagogical motives, the instability of genre now rose into view: the valid work does not merely conform to its genre but modifies it (e.g. Sterne’s Tristram Shandy) or cuts across genres (e.g. Joyce’s Ulysses). In this outlook, the genre is continually in the process of being constituted and reconstituted and is therefore more a part of the problem of classifying literature than the solution.

The concept of style as ornamentation moves toward the concept of style as a mode of literariness, one mode of discursivity among many others, but a highly influential one in some cultures. Like genre, a style requires a dialectic between innovation and expectation in order to assume a distinctive identity, and thus also functions more as problem than solution, especially when an author (like Hemingway) becomes fixated in the public estimation and forced to perpetuate an early style, or conversely, when an author (like the Shelley of The Cenci) abruptly repudiates his own style in seeking out another genre.

Consistent with these shifts, the project of placing a stable value on the text evolves into an increasing sensibility of the transient value of the engagement with the text, a process that unceasingly raises the problem of value without enabling us to resolve it in any enduring or large-scale manner. We may value a work when it innovates or when it meets our expectations, even though neither innovation nor expectedness is a value by itself. We may appreciate of a genre or style without in any way esteeming all of its instances. We may alter our evaluations between different readings of the same work. And so on.

The advocacy of an interpretation and implicitly, also of the text, is less vital now than the advocacy of a model of literary communication. If we are to contain the danger of the old quarrel over the ‘right’ reading of a poem being merely supplanted by a quarrel over the ‘right’ model of poem-readers (see above) -- something of the sort is sporadically and indecisively fought out in the pages of Critical Inquiry or New Literary History -- we need to refine our criteria. Confirmation by eloquence is still quite fashionable, especially when the model itself is as arcane and convoluted as Bloom’s or Hartman’s. The alternative would be confirmation by insight via the modes of literary engagement that the model brings to consciousness; and this is by nature inclusive rather than exclusive.

Also still much in vogue is training by imitation, though in their eagerness to make their models attractive, prominent literary theorists seems paradoxically resolved to bankrupt any potential following in advance with the sheer inimitable brilliance of their performances. Training by method remains in fairly rudimentary stages in comparison, say, to the fieldwork of linguistics and discourse analysis, but is clearly on the advance.

Furthermore the tradition of individual research is still firmly in place, doubtless because the complexity of literary communication seems handiest to master with a personal blend of introspection and demonstration. A group of theorists may reach some critical mass of consensus, as in the ‘Yale school’, but this same group reveals complex variety unravelling uniformity (Bloom and Hartman not being ‘deconstructionists’ in any sense like de Man and Hillis Miller). In most institutions, the prospect of empirical research by teams is still remote or at best marginal. The ‘International Society for the Empirical Study of Literature’, founded in 1987, is mainly a group of sociologists, psychologists, and continental Europeans working outside major literature programs, where their impact on the daily practices of using literature is not likely to be significant.

If the practice-centred tendencies of former times are slowly moving toward a balance of theory with practice, dizzying vacillations can yet be felt. In the first place, the heavy commitment to theory is by no means universally motivated by an intention to transform practice, and in some cases (e.g. Bloom), such an intention is expressly repudiated. In the second place, the practice offered by numerous theorists to demonstrate their models retains a personal touch, and blends together how people read with how they should read. For example, an empirical project designed to test Iser’s model of the ‘implicit reader’ found that ordinary readers do not adjust to ‘gaps’ in the text, but ignore or distort them and cling to whatever does meet their expectations.[20]The ‘implicit reader’ reads so much like Iser himself that we will need extensive bridging if the model is to be widely practicable.

Text Linguistics and Discourse Analysis Compared to Literary Theory

By juxtaposing the shifts categorised in Tables 2 and 3, we can compare the trends in text linguistics and discourse analysis on the one hand and in literary theory on the other, as shown in Table 4. Though several disparities can still be detected here, we by no means discover the stark contrasts indicated in Tables 1, 2, and 3.

Text and discourse are now accepted as central entities on both sides, although theses terms have a wide range of interpretation. For discourse analysis, the emphasis falls squarely on the social and cognitive aspects, while in literary theory, the social ones are emphasised only by the political ‘left’ in cultural anthropology, materialism, feminism, Marxism and so on (e.g. Foucault, Millett, Jameson); and the cognitive ones mainly from the standpoint of phenomenology and gestalt theory (e.g. Iser) and sociology of knowledge (e.g. Jauss, Bleich). Some conspicuous branches of ‘post-structuralism’ promulgate a curiously convoluted and self-directed notion of ‘text’ or ‘discourse’ from which society is essentially absent and cognition is transfixed in ‘aporias’ (e.g. the erstwhile ‘Yale school’). The same limits apply if we compare the view of text producer and text receiver as social and cognitive agents with the models of author and reader, and of literary production and reception. Again, literary theory has not pursued the social and cognitive aspects as extensively and consistently as might be desired to enhance or clarify the design of models or to estimate which of two models is more plausible.

The sources of materials are also somewhat incommensurate, in that text linguistics is in principle concerned with all text types, whereas literary theory is still selective, albeit broadening considerably beyond the traditional ‘canon’, e.g. by addressing ‘trivial’ and ‘popular’ literature. Also, data about production and reception are central to text linguistics but still mainly speculative in literary theory, perhaps for motives of expedience. Fish’s ‘affective stylistics’, for example, is clearly at variance with the findings on real reading in projecting a word-for-word linear procedure and ignoring the hierarchical processing that buffers real readers from the small-scale ‘surprises’ Fish deems so significant.

A particularly marked contrast can be seen between the data-gathering techniques, which, as noted in above, are both empirical and diversified in text linguistics and discourse analysis, but still firmly ensconced within literary institutions for literary theory. The ‘Empirical Society’ sketched in the third section represents a noteworthy counter-trend, but its potential to exert a major impact on those institutions is still very much an open question. [21]

The parallel between a dynamic, procedural orientation and a programmatic one is more compatible, though with some differences in degree and focus. Literary theory has adopted a wealth of programmatic approaches that are certainly more dynamic not merely than traditional literary studies but such early theoretical trends as Formalism and New Criticism. The extent to which they could also be called procedural is less readily evident, since, as I remarked, the striving for eloquence and brilliance tends to obscure the underlying procedures of the actual reading .

The concept of the community as social complex is in principle relevant for the concept of ‘horizon’ in that literary expectations are current among a substratum of that complex. Yet the status of that substratum within the whole is still much less well-defined than would be desirable in view of the widespread but largely planless use of literature in public education (see last section).

The balance of general and specific in text linguistics is only partly compatible with the balance of innovation vs. expectation in literary theory and is otherwise concerned with a whole range intermediary constructs between the whole language and the single text -- text type, register, style, context, situation, and so on -- which form the framework within which anything may be more or less innovative or expected.

The concept of strategy as heuristic would be helpful for appraising both the instability of genre and the diversity of style, since for the text producer, both genre and style are more projects or targets than facts or artefacts. We could thus envision a style or genre as a complex of strategies guiding text production along with the more general cognitive and linguistic strategies addressed by empirical research so far. The two conceptions of style are accordingly quite proximate: neither something in the language or in the text, but a mode of discursivity which may be literary, non-literary, or quasi-literary (e.g. advertising jingles).

If discourse is evaluated by interactional criteria, we can apply the same outlook to ask whether the traditional values in literary studies were relevant or favourable for the interaction of authors and readers; and considerable evidence suggests that they were so only in special circles but not in general education. If values are served up as predecided and bound up with a single ‘correct’ reading, then the reader’s opportunity for a self-reliant value of engagement is abridged or even alienated. Literary theory provides good reason to expect major advantages from encouraging readers to discover their own values and readings, and to acknowledge the dependency of value on the richness of those readings. This project would constitute one application to discursive practices that would also support an advocacy for a model of literary communication as the development and enrichment of the self and the imagination. This would clearly constitute confirmation both by social relevance and by increased insight.

As remarked above, training by method is more advanced in discourse analysis than in literary theory, mainly because of the diverse traditions inherited from general linguistics on the one hand and from literary studies on the other. The degree to which the investigator engages with the method is at present disputed on both sides, though ‘critical engagement’ is rapidly gaining ground.

The prospects are favourable for the reconciliation in research of a self-reflective orientation with an empirical one, but much is still to be done. In discourse analysis, empiricity has tended to discourage self-reflection in some quarters, but the balance has been impressively even-handed in most.[22]In contrast, the intense self-reflections in literary theory so far have seldom led to a genuinely empirical alignment that could, among other things, provide material criteria for the advocacy of models (cf. p. 11).

Finally, the balance of theory with practice is yet a bit uneasy on both sides, but the signs are encouraging, and the isolation of theory from practice so often witnessed in the past is barely admissible today. What remains to be achieved, however, is the alignment with the wider social practices of discourse beyond the bounds of the disciplines themselves, and I shall conclude by briefly examining this point.

Practice as a Social Problematic

Discourse is not merely something that people learn to produce and receive, but something that mediates most other modes of learning. Therefore, the need for application to practice is nowhere more urgent than in the institutions of socialisation and education. The apparent order of ‘curricula’ and ‘lessons plans’ is usually based on a naive, reified categorisation of the subject matter into ‘content’, e.g. history as a batch of people, places, and dates, rather than a social and political evolution following distinctive processes of power versus solidarity.

From the vantage point of discourse analysis, the chief defect in socialisation and education is the failure to appreciate the full role of discursivity as the central mode for accessing knowledge and to draw the consequences. Substantial research, notably at the Centre for the Study of Reading (University of Illinois, Urbana) has been able to demonstrate the diffuse, often inefficient design of such materials as textbooks and lectures. Viewed apart from discourse, facts and figures are all too easily imagined to be straightforward little packets of knowledge waiting to be collected. In discourse, however, they are merely incidental points that become meaningful and memorable only in rich contexts, which the learners have traditionally been left to design on their own; and if they did not prosper at this sophisticated task, they were classed as ‘average learners’, which in effect meant that they should not expect any special consideration from the educational system. This large middle group was treated offhandedly for not being ‘intelligent’ or ‘bright’ in the peculiarly narrow and uncreative senses generated by a system that hoarded facts and figures but could not meaningfully communicate them in relevant contexts.

In the ‘language programs’ of educational institutions (e.g. English Departments in the US), the traditional preoccupation with a certain quasi-literary brand of fastidious (‘correct’) prose all but eclipsed the realistic development of those language skills for which the majority of learners (the ‘average’) already had the prerequisites by virtue of speaking the language.[23] Most uses of literature were either historical (people, places, and dates) or, where interpretation was involved, authoritarian. The total message transmitted to the learners was that they were not fully incompetent to read or write, though, by dint of strenuous exertions, perhaps able to make a reasonably good showing now and then on an essay or test.

The imperative today is clear enough, however arduous and remote its realisation: an integrated, discourse-centred approach to the entire educational experience, placing the language program in the pivotal (and rather daunting) position of training the discourse skills for navigating both in everyday life and in the several domains of schooling itself. For example, geometry could be approached as a ‘special purpose discourse’ about a system of idealised spatial relations,[24] and English teachers would work with the geometry teachers to co-ordinate their own focus. In such an environment, the teacher’s task would no longer be to dispense isolated facts and correct or punish deviations from these, but to serve as expert and consultant within the discourse about integrated domains where a set of facts can become meaningful and instrumental in the production of further knowledge.

The university level is of course the ambience where this design must first be developed, but also where it must first be put into practice. Current programs for ‘writing across the curriculum’ and for the investigation of special-purpose language and terminology are steps in this direction, but a much more sweeping reorganisation and co-ordination will be needed before discursivity attains the pivotal role it merits. Every specialised area needs both explicit coursework by learners in the appropriate discourse and regular reassessment by experts of the prevailing terminology.[25] Here too, the university must develop model educational materials that can support a general reorientation in lower-level schooling.

If my assessments of the recent shifts in the second and third sections seem unduly optimistic or premature, and my projections in the fourth and fifth sections unduly expansive and ambitious, I would respond that more traditional assessments and projections have been slanted in the opposite direction. Consistently perpetuating routines, pursuing mosaics of incidental subtasks, and taking it for granted that theory and practice were aligned as well as they needed to be, has allowed the disciplines based on language to drift ever deeper into a unacknowledged crisis, wherein neither theory nor practice seems adequate to a very conspicuous crisis in global communication arising in the wake of the information explosion. Moreover, the continuation of ineffectual or authoritarian routines saves labour at the upper end (teacher or expert) only to waste it at the lower end (learner or novice). It has thus not been realised that the substantial proportion of ‘average’ result or even of failure in socialisation and education is due not to low intelligence, lack of aptitude, or laziness, but to fundamentally unbalanced and ineffectual discourse, and is therefore not a natural product of a competitive system but a egregious denial of the personal freedoms guaranteed in principle by modern democracies. For everyone’s sake, including ours, we must henceforth do all we can to bring about freedom of access to knowledge, in and through discourse.

Notes

[1] This paper grew from an invited seminar at the University of Southern Florida in Tampa.

[2] Around 1880, linguistics became distinct from philology through the work of major scholars like William Dwight Whitney, Jan Ignacy Niecislaw Baudouin de Courtenay, Mikolai Habdanc Kruszewski, and Henry Sweet; and the study of modern language and literature separated off from ancient through the creation of actual university chairs -- a development for which linguistics would still have to wait quite a while.

[3] Quoted from Chomsky (1965:9). The widespread misreading of the term in the sense of ‘produce’ was probably desired, e.g., to shore up the thin claims that generative grammar captured the ‘creativity’ of language. Chomsky (1965:8; 1957:17) offered ‘technical devices for expressing a system of recursive processes’ as ‘an explicit formulation of creative processes’. Yet recursion is the exact opposite of creation and just churns out the same thing at fixed increments. The real creativity of language, as shown, say, in poetry, falls in a major trouble zone of generative theory, namely on the borders of the ‘grammatical’.

[4] Aside from Peter Hartmann, few general linguists or text linguists appear to have foreseen this development.

[5] Clear evidence for his convergence can be seen in the consensus of authors in the 10th-Anniversary issue of the journal Text (van Dijk [ed.] 1990).

[6] See Beaugrande (1980, 1984).

[7] For surveys, see Beaugrande (1980-81, 1982, 1984, 1997); Heinemann & Viehweger (1991).

[8] Chomsky (1965: 3) attributes this ‘idealisation’ to ‘the founders of modern general linguistics’, but I could not find anything like it in the writings of Saussure, Sapir, Bloomfield, Firth, or Hartmann, who surely count as founders. At most, Hjelmslev contemplated ‘eliminating’ ‘accidents' and `’disturbances’ ‘in the exercise of language’ (in ‘parole’) (1969[1943]: 94), but he had no ‘speaker’ or ‘hearer’ at all.

[9] On this distinction, see especially van Dijk and Kintsch (1983).

[10] See for instance Chilton (ed.) (1985); Wodak (ed.) (1989); van Dijk (ed.) (1990); Fowler (1991).

[11] For an application of this proposal to general linguistics, see Beaugrande, Theory (1991).

[12] See for example Riley, Greeno, and Heller (1982); van Dijk and Kintsch (1983).

[13] E.g. Coulthard (1985) (linguistic) and Macdonnel (1986) (literary).

[14] For an astute critique of this notion and an impressive project directed by Herbert Simon for modelling scientific creativity through simulation, see Langley et al. (1987).

[15] The exact translation of the original German title of Iser’s (1972) book, whereas the ‘implied reader’ in the English version (1975) goes further in suggesting an act of projecting.

[16] Indeed, Wellek and Warren had listed Bloom and Hartman as model practitioners of close reading!

[17] For Rousseau, see de Man (1981), and for Shelley, Bloom et al. (1979).

[18] Compare the papers in the first issue of volume 18 of Poetics (1989).

[19] Seminal papers collected in Warning (ed.) (1975).

[20] See Hömberg & Rossbacher (1977). Oddly, Iser told me he considered this study a confirmation of his model.

[21] Or indeed its intention to do so. A resolution submitted by myself in collaboration with S.J. Schmidt and Gerhard Rusch at the first general meeting calling for the support of emancipatory uses of literature was violently attacked by the planned organisers of the second general meeting and tabled without a vote.

[23] See for example Atkinson & Heritage (eds.). (1984); Drew & Heritage (eds.) (1992).

[23] For a basic textbook designed to work on that basis, see Beaugrande (1985).

[24] A pilot study for geometry is developed in Beaugrande , “Knowledge” (1991). Papert’s (1980) excellent LOGO project for learning mathematics by computer simulation does not recognise the role of discourse, probably because the child is mainly self-communicating by writing simple programs on a terminal.

[25] See Beaugrande, “Communication” (1991).

Works Cited

Atkinson, J.M., and J. Heritage Eds. Structures of Social Action: Studies in Conversational Analysis. Cambridge: Cambridge UP, 1984.

Baudouin de Courtenay, J. Otryvki iz lektsii po fonetikie i morfologii russkogo jazyka. Kazan: Kazanskom Universitet, 1882.

Beaugrande, R. de. Text, Discourse, and Process: Toward a Multidisciplinary Science of Texts. Norwood, N.J.: Ablex, 1980.

Beaugrande, R. de. “Design criteria for process models of reading.” Reading Research Quarterly 16 (1980-81): 261-315.

Beaugrande, R. de. "General constraints on process models of language." In Language and Comprehension. Ed. J.-F. le Ny and W. Kintsch. Amsterdam: North Holland, 1982. 179-192.

Beaugrande, R. de. Text Production. Norwood, N.J.: Ablex, 1984

Beaugrande, R. de. Writing Step by Step. New York: Harcourt Brace Jovanovich 1985.

Beaugrande, R. de. Critical Discourse: A Survey of Contemporary Literary Theorists. Norwood, N.J.: Ablex, 1988.

Beaugrande, R. de. “Toward the empirical study of literature: A synoptic sketch of a new society.” Poetics 18 (1989): 7-27.

Beaugrande, R. de. Linguistic Theory: The Discourse of Fundamental Works. London: Longman, 1991.

Beaugrande, R. de. “Knowledge and discourse in geometry: Intuition, experience, logic." Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 6 (1991): 771-827.

Beaugrande, R. de. “Communication and freedom of access to knowledge as an agenda for the special purpose language movement.” Fachsprache 10 (1991).

Beaugrande, R. de.New Foundations for a Science of Text and Discourse: Cognition, Communication, and the Freedom of Access to Knowledge and Society. Greenwich, CT: Ablex, 1997.

Bleich, D. Subjective Criticism. Baltimore: Johns Hopkins UP, 1978.

Bloom, H. Anxiety and Influence: A Theory of Poetry. New York: Oxford UP, 1973.

Bloom, H. A Map of Misreading. New York: Oxford UP, 1975.

Bloom, H. et al. Deconstruction and Criticism. New York: Seabury, 1979.

Bloomfield, L. Language. New York: Holt, 1933.

Chilton, P. Ed. . Language and the Nuclear Arms Debate. London: Pinter, 1985.

Chomsky, N. Syntactic Structures. The Hague: Mouton, 1957.

Chomsky, N.