TEXT PRODUCTION

Toward a Science of Composition

Robert de Beaugrande

University of Florida

1984

Volume XI in the Series

ADVANCES IN DISCOURSE PROCESSES

Roy O. Freedle, Editor

A Retrospective from 2004

    My original impetus to undertake this book came from being, yes, an English composition teacher, a calling widely regarded (even by others in it) as a lowly drudgery mostly for pursuing and pillaging the “mechanics” (“misspellings”, “bad grammar”, “sentence fragments”, and such evils), sullenly awaiting the apotheosis up to ‘teaching high literature’. To my surprise and dismay, many of my colleagues saw no special need to inform themselves about their profession; they hadn’t even read such basic books as Ed Corbett. Mina Shaughnessy, or Gary Tate’s fine gallery of “bibliographies”. They ‘taught’ by intuition, by personal notions and biases, often just copying their own previous composition teachers, ignoring the dramatic demographic changes of language varieties among students under way then.

    When I became involved on the Conference on College Composition and Communication (CCCC, spoken as ‘the four C’s’), I seemed to sense a sluggish generation shift of power from the old mandarins who saw no special significance or revelevance in what I would call ‘empirical research’ (like Ross Winterowd, Ed Corbett, Frank D’Angelo, Joe Williams, Willliam Irmscher, and (gag) Frank O’Hare) — plus, abruptly and belatedly , Don Hirsch — and a younger crowd who were active in specific areas of research (like Mike Rose, Andrew Kerek, John Daly, Martin Nystrand, Dixie Goswami, Ann Feldman Matsuhashi, Robert Gundlach, and  Stanford Gwim), plus a loose collage of bright minds who were not in the composition field per se, but had some related interest in writing or text production (like Frieda Goldman-Eisler, Wallace Chafe, Carl Frederiksen, Michael Halliday and Ruqaiya Hasan, Carl Bereiter and Marlene Scardamalia, Herbert and Eve Clark, James Deese, Walter Kintsch, Heinz Mandl, Olga Akhmanova, and Jozef Vachek).

    What I did not see and what prompted me to write this erm, large book, was the lack of work I could find on text production from a fully multidisciplinary perspective on its vast but somehow integral complexity. So I plunged in headfirst, with unflagging encouragement of the late and lamented Walter Johnson, founder of Ablex — but not from “research foundations”, one of whom who actually wrote “The Principal Investigator is in an English Department and therefore [my italics] can have no deeper understanding of these concepts” — gormlessly unaware that I had introduced some of those concepts in earlier years).

     I did not anticipate such a mammoth ‘research work’ with 1242 ‘References’ (which still has far too many omissions), but, well, there it was, and Walter (then in his 80s, but “with the spirit o' five-and-twenty in him still”, as Sam Veller would attest) said “whatever you like”, saving me from the routine jeremiads of arrant “reviewers” whose very job is to know better than I do how to write books — especially fat books which further annoyed them by having no model or precedent. I think nobody every reviewed it either, probably for the same reason.

      As far as I could tell, it was (to put it mildly) not joyously welcomed by the CCCC, most of whose members, mandarins or not, stayed in their safe, well-paid corners. But it did reach quite a few other disciplines, often outside the US of A. including the ‘socialist’ countries, where it was passed around very surreptitiously, as I later learned.   

    I should stress here that this and my later books are NOT rehashes of my earlier ones, in contrast to some authors I won’t name here (however much they deserve it). Most of my larger schemes and terminologies were, at the time, I believe, original, such as the principles of linear systems. Neither is it an arcane ‘theory’ book, as many called it as a pretext to stick with little ‘practice books’ like (yawn) Strunk and White). It is, on the contrary, filled with actual teaching techniques, pilot projects with results, and student data. And 550 text  examples not made up by me., and mostly authentic student data.

     And then there are touches of humour for those who relish them, like Frank O’Hare’s floating rock formations (II.3.45). Science should be fun, too.  

Technicalities of the Format of this E-Version  

   I have occasionally improved on the original wording, but of course have not tried to convert the viewpoint from 1984 to 2004. In return, I was very pleased that not much is said here that I would repudiate today. The ms. was written on an Apple II (with those ghastly 5 ½ floppies that held about five pages each); and the graphics, which I did by hand in those days, are now jpgsm and  even after I did some arduous retouching, are not as dark or clear as I would like. The font is a user-friendly 12-pt Ariel, with  the150% line spacing which the Internet seems to demand. Both font and spacing could be reduced if you wanted to make a print-out., say 11pt Times New Roman, single spaced.

     Since page length was increased to conserve bulk, the original pagination is not shown as such. Numbers in {curly brackets} mark the old page divisions, but I omitted them where they might have been intrusive or confusing;  write me if you need any you don’t find.   Footnotes have been moved to appear in [square brackets] right after where they were marked..  The paragraphs are numbered for easier cross references than by page, though I have heard colleagues say ‘”I won’t read a book with numbered paragraphs,” which they find "intimidating". One wonders if they read at all.

CONTENTS

I. The Context of Communication

1. A “science” of composition?

2. Literacy: Critique of a crisis

3. Evaluation and grading

4. Text and discourse as events

II. Scientific Precursors

1. Structuralism and linguistics

2, Physicalism and behaviorism

3. Mentalism and transformational grammar

III. The Procedural Approach

1. Designing a model

2. The phases of text production

3. The resources of text processing

IV. The Linearity of Text Production

1. Temporality versus spatiality

2. Seven principles of linearity

3. Linearity and the motives for punctuating

V. Writing vs. Spelling, Grammar, and Speech

1.  Spelling: Rules, regularities, or randomness?

2.  Grammar made operations

3.    Speaking versus writing

VI. Style, Stages, and Steps Toward Progress

1.   Style

2.   The stages of discourse

3.    Theory and practice: A common cause

References

The Context of Communication  

 A “SCIENCE” OF COMPOSITION?

  1.1 A “science” of composition may seem an unusual notion. Science is widely considered a specialized enterprise for making statements about stable entities and quantities. Webster’s well-known dictionary (1963: 771) defines “science” as “knowledge covering general truths, the operation of general laws,” and a sub-definition is: “one of the natural sciences” — the latter “dealing with objectively measurable phenomena” (1963: 653). These definitions embody the common belief that any domain hoping to be considered a legitimate “science” should emulate the natural sciences and their conceptions: laws, measurements, generality, and objectivity.

 1.2 Early science was rooted in the speculations and ideologies of philosophy, theology, and politics. Facts were established by assertion (cf. I.2.8.4), and accepted as an act of faith in the person or institution making the claim. Modern science strove toward a method where facts are established by empirical discovery and accepted as a conclusion drawn from evidence. However, these strivings have had some problematic side-effects. Scientists became skeptical about the responsibility of science to serve the interests of society at all. Research focused on direct manipulation of small scale objects in the laboratory and discounted their contexts in everyday life. Statements were mainly quantitative, rather than qualitative, because numbers allow neutral comparisons and generalizations more easily than characteristics do (cf. II..2.2). Individual objects were leveled into a statistic norm: the highest truth was equated with whatever occurs most often. In principle, however, the norm may be a theoretical construct not embodied in any one real object.

 1.3 In this ambience, the human sciences looked for domains that seemed well-structured. If necessary, domains were simplified and broken down via reduction and fragmentation. Static objects were stressed at the expense of dynamic events. Hence, scientists preferred to study artifacts rather than the processes whereby those artifacts are produced and used within human activities; and to treat orderly special cases as general cases or as metaphors for the latter. Contexts and processes — especially those involving human creativity — were widely believed to lack the orderly structure and uniformity required for valid scientific inquiry. In actual practice, the investigators themselves informally supplied these contexts and processes. Methods worked best when the artifacts could readily be fit into uniform contexts (cf. I.1.7). For example, linguistics analyzed the formal structure of words and sentences much better than that of meanings and purposes (cf. II.1; II.3).

 1.4 After a time, research of this kind must reach a point of diminishing returns, because a major part of the inquiry is hidden from view and thus hard to control. As long as scientists supply contexts without acknowledgement, the treatment of facts and evidence contains a concealed intermediate step that may endanger the consistency and reliability of the findings. The ideological and personal biases modern science strives to exclude might creep in at this very point. Such was the fate of the three scientific trends to be surveyed in Chapter II: structuralism, behaviorism, and mentalism. By discounting or fragmenting ordinary human processing, they failed to get any general picture of communication and cognition. This impasse can be resolved only if we agree that scientific principles are not compromised by seeking structures in processes, not just in artifacts. We can be more certain of what we know if we explore how we come to know it (I.1.8; III.1.32).

 1.5 Accordingly, the status of science has recently been re-examined along three dimensions: historical, philosophical, and psychological. Historical studies (e.g. Kuhn, 1970a) remind us how, in the past, ardently accepted facts collapsed despite the methodologies and institutions that favored them. Conversely, many facts we now take for granted once encountered protracted opposition from high-ranking scientists. The border-line between a “new discovery” and a “scientific mistake” is precariously negotiated. Science must be carefully maneuvered between the tendencies to explore and to ignore. If we bear in mind that any set of facts is in principle open to more than one explanation, and that no truth can be final, we will be more willing to consider new or competing approaches.

 1.6 Philosophical studies lead to a similar conclusion via a different route. A fundamental principle is currently in debate: the verification and falsification of theories (cf. Hanson, 1958; Feyerabend, 1970a; Lakatos, 1970; Popper, 1970, 1972; Stegmüller, 1976). The uncertainty of the dividing line between evidence and the interpretation of evidence has been recognized. Prevailing theory creates the categories in which scientists perceive and think, and thus decides what counts as evidence; a wholly impartial judgement is epistemologically not feasible. Contrary evidence that cannot be conceptualized within customary approaches tends to go unnoticed. Scholars who advance these considerations (e.g. Feyerabend, 1970b; Kuhn, 1970a, 1970b) have been attacked by defenders of the traditional status of science for portraying scientific communities as “irrational”¹ (cf. account in Stegmüller, 1976) [¹To the advocates of one theory, those of another naturally appear irrational, because coherence is being derived from two different frameworks. Creativity always entails a momentary disturbance of thought patterns, but is not irrational because it engenders a new order (cf. I.1.3; I.2.23.7; 1.3.7, 10)., i.e., as insensitive to contradictions and inconsistencies.² [For example, a vision program for a computer must incorporate knowledge about physical objects (dimension, location, displacement, edges, shadows, etc.) (cf. Winston [Ed.], 1975). Otherwise, the world looks like a jumble of lines and intensities capable of staggering numbers of interpretations.]  But a semblance of irrationality is a natural by-product of the scientific enterprise. As theories and explanations change or evolve, inconsistent or contradictory sets of facts sometimes persist side by side. Indeed, science deliberately fosters such a state of affairs by seeking out problems and anomalies. Scientists understandably maintain their allegiances at least until a new corpus of contrary evidence becomes relatively large and secure. The real dispute is how this evidence weighs against prior institutional or personal commitments, e.g. one’s investment in the older theory.

 1.7 Psychological studies endow the message of the historical and philosophical ones with a new urgency. Researchers concur that “in every domain of human experience, perception, comprehension, and interpretation involve an interaction of input with existing knowledge” (Anderson, 1977: 417). Everything humans can discover or know, except perhaps the experiences of very young children (cf. Piaget, 1976)¹, makes sense only via the mediation of our prior model of the world. What is true of humans at large is perforce true of scientists: there can be no absolute objectivity because there can be no totally unmediated access to an object. People agree about an object (and consider it objectively real) when they attain consistent and comparable processing results from encountering or using the object. Processing is therefore not chaotic; its profoundly ordered structuring helps to build the order of the world.

 1.8 The most pressing issue for science now is understanding understanding (cf. Rumelhart, 1981). Until we obtain a viable model of what it means to understand something, the credentials of science itself have no ultimate recourse.¹ [ Neisser (1982: 44) observes that psychoanalysis and behaviorism imply a self -contradiction because they would exempt themselves from human nature. “If all ideas are wish-fulfillment, then psychoanalytic ideas just express the wishes of the psychoanalyst. If every belief just results from conditioning, so does the behaviorist’s belief that this is the case.” Cognitive psychology, in contrast, is concerned with “how knowledge is possible,” and thus can deal with itself as one domain of human nature, rather than claiming to be a miraculous exception to its own scheme of things.] All essential constructs of science — discovery, objectivity, measurement, generalization, proof, etc. — are human processing acts we need to expound and explore (Neisser, 1982). The procedures of science should be analyzed in their interaction with commonsense reasoning (cf. Garfinkel, 1960; Cicourel, 1964; Jennings & Jennings, 1974). We can come closer to reality by examining the procedures whereby we constitute reality (I.1.3).

 1.9. We will do well to take seriously Tulving’s (1979: 29ff) admission that “all current ideas, interpretations, and theories are wrong, in the sense that sooner or later they will be modified or rejected.” Such a tenet need not damage our initiative and self-image. It might, on the contrary, incite us to contemplate unmapped terrains of discovery. And the imperative to question our answers (and even our questions) keeps us from failing into rigid habits of thinking. Tulving (1979: 29ff) himself draws some noteworthy consequences (cf. VI.3.19-28):

 1.9.1 It is unwise to expend too much effort on affirming accepted theories. Experiments should be designed not simply to shore up such theories, but also to show how they are still incomplete or inadequate.

 1.9.2 An account for the findings of a single experiment or type of experiment is not too valuable. We need models that include the larger context in which any activity is performed. Such models “will bring out relationships between experiments or sets of data that we would not otherwise have perceived” (Estes, 1975a: 271).

 1.9.3 Formalization should not be undertaken too early. Unwieldy constructs borrowed from mathematics and logic are out of place in domains where the basic concepts are still highly approximative. Such constructs give a false sense of security of having explained what has in fact only been rewritten in a formal language.

 1.9.4 Scientists should be more willing to reject accepted theories that no longer lead to new or enlightening findings. The exertions of repeatedly patching up faltering theories could be better invested in the search for new ones.

 1.10 The human sciences have both as their own foundation and as their object of inquiry the procedures of MODEL-BUILDING (cf. Beaugrande, 1980a, 1981a, 1981b). A MODEL can be defined as a theoretical  construct intended to represent the nature and workings of some object or object domain. As long as the domain is relatively unfamiliar, the model remains highly APPROXIMATIVE, i.e., contains provisional assumptions for which there is sparse evidence. Further experience and reflection can TUNE the model, i.e. bring it into steadily more reliable correspondence with the domain and provide more consistent, effectual orientation. A society or culture is defined both by its major models and by its customary procedures for building and tuning them. At one extreme, the model becomes totally internalized and invisible, seeming to be reality itself; at the other, people become aware of models inside models (build models of themselves building models, and so on indefinitely), in an infinite regress.¹ [¹”Infinite regress” is the philosophical and psychological consequence of the search for ultimate causality and definition. In everyday life, people set a threshold where they stop pursuing the causes or definitions of one artifact or operation in another, and accept things as given. In science, the hope lingers on that physiology and neurology might be the final cause of cognition and communication (cf. II.2.3ff). But since the causalities are discoverable only in terms of a complex process model, the latter cannot be strictly derived from physiological events and observations.] Either way, models remain pervasive and powerful determiners of how the world is viewed.

 1.11 A science will progress the most swiftly by remaining aware of its own model-building. Science has special demands for orientation: not only to experience a domain, but also to describe and explain it. A scientific community typically seeks high degrees of completeness, consistency, and exactness, whence the familiar conceptual apparatus — definitions, classifications, measurements, regularities, and the rest. There is no way to conclusively verify a model, because all possible evidence can never be assembled (cf. Popper, 1972); at most, the model can be falsified if it contradicts a significant body of evidence (cf. Lakatos, 1970). However, the standard methods of observing, experimenting, predicting, etc., are never completely free of the prior model that determines which answers are sought for which questions. The real danger is that constant application of a model can internalize and reinforce it to the point where it is considered identical with the domain itself. At that point, the model severely limits and obscures the negotiation of the threshold at which accruing contradictions overturn an established theory (cf. I.1.5).

 1.12 The priorities proposed in 1.9.2 suggest a way out of this basic predicament. The model can be DIVERSIFIED by applying it to a steadily wider domain or set of domains. Each step of diversification reexamines the accountability of the model. Of course, the model still influences discovery procedures and leads scientists to incorporate their prior assumptions into their treatment of evidence. But these effects are structurally and functionally altered by each step of diversification. The model must account not for just one concern, but for a growing intersection of  concerns. The probability steadily rises that an inadequately designed model will lose its plausibility. Diversification should be based on ANTECEDENCE: the procedure of progressively specifying a general theory to create sub-theories for several domains. Research in any one of these domains can then be related to the others via an explicit shared conceptual framework. For example, a theory of communication and cognition is properly antecedent to a theory of text production, and the latter in turn antecedent to a theory of writing. In contrast, EXTRAPOLATION entails borrowing notions, methods, accounts, etc., from a “source” domain and transferring them to a “goal” domain without any such overarching framework. Normally, extrapolation occurs because the concepts and procedures of the source domain are better defined and operationalized than those of the goal domain. On occasion, the unexploredness of the goal domain can preclude a critical evaluation of whether the transfer is theoretically justifiable, and not just convenient. There may be no more than a metaphoric similarity between the domains, so that the well-defined source domain says very little about the precise nature of the goal domain (cf. I.1.3; II.1.2.15, 24; II.3.9; III.2.1). 1.13 The context for a “science of composition” should therefore be established through antecedence. As a sub-domain in a theory of text production (including speaking), composition is necessarily subject to the general conditions of cognition and communication. Texts are produced through complex operations that are guided by available processing resources, such as memory, attention, motor control, feedback, and motivation. These conditions in turn affect the practical applications through which theory and research on text production fulfills its responsibility to society (cf. VI.3). By exploring situations where text production is a relevant human action, we can better support the skills of communication in general and of literacy in particular. As a social institution, literacy promotes complex model-building, both collective (social, historical, scientific, technological, etc.) and individual (introspective, immediate, etc.). Scientists and writers alike use approximations and tests to steadily narrow down the disparity between intention vs. action, knowledge vs. expression. Thus, understanding the origin and production of discourse can bring new foundations not only for language instruction, but also for a general epistemology of discovering and communicating knowledge.

 1. 14 The importance of establishing antecedence relationships among adjacent disciplines is now widely recognized. The fragmentation of theory and research so common in the past (I.1.3) is yielding to unification and interdisciplinarity. Still, the legacy of previous research can serve us in the larger enterprises of the future. Chapter II surveys three scientific precursors that approached language or language activities via some degree of extrapolation. Structuralism was based on the notion of analysing objects into minimal components, as practised in the natural sciences (II. 1). {7} Behaviorism centered upon the notion of conditioning, as established in animal testing (II.2). Mentalism centered on the notion of axiomatic systems, as conceived in formal logic (II.3). In each case, the application of these notions to human language was sometimes contestable, reductive, or misleading. However, many of the issues then raised still remain valid for any theory of language or communication, provided we find a unifying mode of stating and pursuing them.

 1.15 Though composition itself is cannot be a science, it constitutes a rich domain for scientific inquiry into human processing. This volume is intended to outline a provisional research program such a science might profitably pursue. In so early a state, approximations and limitations cannot be avoided (cf. I.4.17; VI.3.25). For the time being, qualitative discovery seems more vital than quantitative (cf. I.1.2; 1. I.9.3; I.4.17; VI.3.22). We need to clarify the terrain before exact measurements and formalizations are likely to become decisive. We also need to determine how the detailed factors of text processing can be classified and described in relation to the whole. Finally, we should not forget that the conviction carried by a theory or a model ultimately depends on what Ulric Neisser (1976: 7f) calls its ECOLOGICAL VALIDITY (III.1.31.8): cognition and communication as they “occur in the ordinary environment and in the context of natural purposeful activity!”  

2. LITERACY: CRITIQUE OF A CRISIS

  2.1 In spite of educational training in schools and professions, only a small portion of human abilities and resources are tapped in everyday life. Traditionally, the learners themselves, rather than the educational approaches, have been blamed for limitations and failures. The learners, it was argued, were inherently incapable of better performance, due to factors like “genetic endowment” and “intelligence quotient” (I.3.2). Quite conceivably, however, a fundamentally different educational approach could greatly reduce the ratio of failure schools now accept as normal and necessary (cf. Combs, 1979; Marks, 1982). Barring cases of actual brain damage, every child could acquirefrom schooling a sufficient level of knowledge and skill to understand and control his or her natural life (cf. Bruner, 1966). All we have proven so far is that current schooling does not attain this result.

 2.2 The so-called “literacy crisis” may thus be a result of educational policy, not just a barrier to the latter’s projects. On the one hand, the crisis is illusory, resulting less from a decline in skills than from a profound change in the fabric of society and its cultural expectations. Educational opportunities have been extended for the first time to disadvantaged social groups who contribute a nontraditional population of learners (Roeche & Kirk, 1973; Cross, 1974). These students don’t share the hidden preconditions built into curricular domains such as literacy. Formerly, English teachers {8} “assumed a cultural trust, a vast body of unspoken but shared routines and information, which freed them from the need to explain what they were up to” (Shaughnessy, 1976: 153). Now that students are no longer attuned to the older goals and methods, teachers readily misconstrue the emergence of diverse backgrounds and dialects as a sudden decline in student potential.

 2.3 On the other hand, the crisis is genuine to the extent that current patterns of social activity in America include literate concerns less prominently than before. By temporarily isolating a person from the group, reading and writing may engender anxieties (cf. I.2.10). Participation in mass media, despite a widespread lack of intellectual challenge, seems to promise social integration and divert attention from one’s own problems. Reading and writing are essentially active and creative, whereas mass media encourage passive, reactive attitudes. The latter experiences could leave one’s processing capacities underdeveloped, so that literate activities would demand an uncomfortable strain. For example, the camera perspective and sound track of filmic media pre-empt the tasks of deciding what is noteworthy or significant in the totality of a scene or event. The visual and acoustic impact of a film can overpower its conceptual organization.¹ [Marshall McLuhan (personal communication) suggested that media impact may alter cognition so radically that literacy is becoming no longer feasible. But his argument, based on the dominance of brain hemispheres (cf. II.2.3ff), failed to consider the adaptability of cognition to new conditions.] Carolyn Stopher, one of my students at the University of Florida, interviewed her dorm-mates immediately after a TV news show. They could report vivid sights and sounds, but couldn’t give any coherent account of whole events and their significance.

 2.4 In another, less publicized, sense, the literacy crisis is genuine because a literacy deficit is a serious obstacle to personal and social advancement. Few people can dispense with literate skills and still have equal opportunities for a successful career in schooling and in later life. A literacy failure is not just a low grade on somebody’s curve, a numerical abstraction to be shrugged off by teacher and pupil alike. It is a tragic waste of human potential to be combatted with every resource available to our social institutions of development and advancement. The denial of literacy is a denial of freedom.² Historical evidence for this denial is the fact that teaching slaves to read or write was punishable by law in the American South before the Civil War.]

 2.5 Just as I do not believe educational failure is either normal or necessary (I.2.1), I cannot take the failure statistics of literacy education as proof that the skills involved are simply too hard for the contemporary learner (cf. I.2.23 1). Appropriately designed methods can provide a workable level of literacy to many children now denied that privilege. The key is to understand, expound, and train the requisite skills, instead of just assuming that {9} they are obvious to everybody with the right social background (cf. I.2.2). Failure should be limited to cases where the learner actively opposes the educational process and refuses to participate. I suspect that such a defiant stance is likely only when learners emphatically doubt that a method can lead to success. If a task is explicit and realistic, learners will exert themselves far more intensely than if it is vague and elusive (I.3.27). The latter description often applies to the task of writing as posed in schools and colleges (I.2.17).

 2.6 Non-traditional learners frequently suffer severe neglect of their literacy skills throughout their schooling. If prevailing methods presuppose a literate background, rather than being effectually able to instill it, then non-literate students are hardly likely to make much headway. Lacking an exact notion of the human processes that constitute literacy skills, educators are badly equipped to observe a child’s progress and diagnose or relieve ongoing obstacles. Somewhere during the developmental process, a learner may adopt a non-strategic language model that blocks further evolution (III.1.8), and may emerge barely literate after twelve years of public schooling. To prevent such an impasse, a science of composition must define those processes and skills clearly enough to provide a more generally applicable approach (V1.3).

 2.7 Sociologically, literacy is a key factor in a whole complex of personal attitudes about human status and values. People are hastily judged by any evidence they present for their literate abilities. Many social and professional opportunities are routinely made difficult or unavailable to people who elicit negative judgements. A particular dialect can mean that a child is expected to fail in school; or that an adult will not be trusted to perform properly on a responsible job. Otherwise enlightened citizens, even educators, misconstrue variations in literacy as marks of low intelligence and ability (I.2.17, 20).¹ [Tom Wolfe (1969a: 117) remarked of a young woman: “But underneath all the ‘gits’ and ‘gonna’s’ she turned out to be probably the brightest girl around there” — as if dialect should normally accompany low intelligence. A survey my student Gerald Langford conducted among rural Florida English teachers showed how they are led by the frequency of dialect-related errors in punctuation, spelling, or sentence grammar to predict that Black children whose home dialects are largely oral will be low achievers.

 2.8 The first step is to analyze the factors that influence literacy judgements. Figure I offers a provisional classification of the parameters along which a person might be rated.  

   2.8.1 Usage: Standard vs. non-standard. The most popular evidence for literacy ratings comes from variations in grammar, vocabulary, spelling, and punctuation. Simple errors, e.g. misspellings, are classed together with well-known markers of non-prestigious dialects, e.g. unusual noun plurals or verb tenses. Self-proclaimed language authorities, with no credentials in serious research, assert the “correctness” of their own usage and aggressively denounce anyone who disagrees (cf. I.2.1 1).

   2.8.2 Access: active vs. passive. “Access” decides the correlation between one’s skills or knowledge and one’s actual performance (cf. I.3.4, 22, 25; II.2.10., III.1.19; III.3.2.2; IV.1.12, IV.2 48; V.1.14f, 36, 40, 48; V2.16; V.3.3, 22ff). Active skills are applied on one’s own initiative; passive skills require activation from outside and thus, depending on conditions, may not be manifested in performance. The psychology of memory has shown that active recall can handle much less than passive recognition (III.3.5.4). Converting passive to active access is a pressing but poorly explored issue in developmental education, e.g. how to bring words from passive to active vocabulary.

2.8.3 Fluency: Fluent vs. hesitant. “Fluency” can denote the rate at which a task is carried out and the smoothness of transitions between successive steps (cf. II.2.5, 28f; II.3.21, 38; IV.2.14, 29, 32, 34; VI.3.12). In spoken text production, fluency depends somewhat on articulateness, i.e. case of producing spoken sounds; but deeper factors like fluent planning and ideation also contribute (cf. III.2.6ff; IV2.14-34). Fluency supports efficiency: the processing of the current text with the least amount of effort (I.4. 10). Fluency may be crucial for personal goals: Wodak (1 980) observed that Viennese courtroom judges, keenly sensitive to the literacy of defendants, believe a fluent report of an incident far more readily than a hesitant one. {11}

 2.8.4 Effectiveness: Success vs. failure. “Effectiveness” can designate the extent to which a text contributes to one’s chances of obtaining a discourse goal (I.4.10). One contributor to effective text production is assertiveness, i.e. presenting one’s materials as true, authoritative, and safe from challenge (cf. I.1.2; 1.2.12, 16.3; VI.1.1; VI.3.25). Another is audience sensitivity: awareness of how text receivers will react (cf. III.1.5; III.2.13; IV2.19.15ff; VI.I.17; VI.2.1; VI.3.12).                            

 2.8.5 Adaptability: Flexible vs. rigid. “Adaptability” is the extent to which text production is tailored to the specific setting and audience. Audiences notice particular style markers (V1.1.4) and respond negatively if the style seems too plain (commonplace, intimate, slangy, etc.) or too elaborated (rare, remote, elevated, etc.) for the occasion. Many items of usage are neutral enough to escape notice, e.g., Quirk, Greenbaum, Leech, and Svartvik’s (1972: 24) example, ‘The student’s work is now much better and seems likely to go on improving’.1 [1. I place all linguistic samples in single quotes.]

 2.8.6 Voice: Unique vs. average. “Voice” can denote the extent to which text production reveals the peculiar individuality of the producer, as opposed to the habits of the whole language community (cf. I.2.16.6; 1.4.8; II.2.28; II.3.44; III.3.10; VI.I.7, 9; VI.3.13). Though each text is normally unique in itself, there are differences of degree and markedness. A text producer with a distinctive voice typically requires from audiences a perceptibly high processing effort by frequently overturning common expectations.
 2.8.7 Participation: Voluntary vs. compulsory. Participation in literate activities may depend entirely on free choice, e.g. reading a novel for entertainment or writing poetry to express oneself. Or, people may be forced to participate against their will, e.g., in some school settings. In between is the range where literate activities are instrumental for some further goal, as in many professions. In some societies, free participation is more prestigious than instrumental. For example, a government official who reads and writes a large number of letters may not be judged as literate as a literary book club president who actually addresses a lesser volume of written materials.

 2.8.8 Domain knowledge: Expertise vs. ignorance. Simply knowing more about the knowledge domain used as a text topic creates the impression of being more literate. There will be easier access, larger and more precise active vocabulary, and often more assertiveness and fluency. Ignorance of the domain, on the other hand, creates the opposite impression.

 2.9 The eight parameters outlined here, though doubtless not complete, should suggest the diversity and complexity of judging literacy. Many of the skills entailed are not just concerns of language, nor are they open to conscious control during text production. Access hinges on capacities for{12} organizing and searching memory, which are hard processes to direct (cf. III.2.24; III.3.10ff; IV.2.11, 27). Fluency depends in part upon individual or biological rhythms of the organism (cf. IV.1.4; IV.2.19.7). Effectiveness is influenced by social beliefs about what goals are desirable (cf. III.2.7). Adaptability is determined by one’s background of experience and practice. Participation may be sporadic or accidental. Knowledge or ignorance of a domain is an accident of one’s training. Thus, literacy judgements have a complex origin utterly belied by the hasty, superficial ways they are often formed.

 2.10 This complexity has two important consequences. The first is that many people suffer from language anxiety because they feel they are being judged by complicated, unexplained standards. Within a very limited range, anxiety is productive in heightening attention and effort; after that, it heavily drains the resources that should properly go to the task itself (Murray, 1971). Anxiety makes the situation appear “difficult, challenging, and threatening”; “the individual sees himself as ineffective, or inadequate, in handling the task,” and “focuses on undesirable consequences” (Sarason, 1980: 6). Hence, anxiety interferes with performance and brings about the negative results the person fears and expects. Anxious attitudes toward literacy commonly impede learning and lead students to predict and accept failure as the natural outcome.^{1 [}Daly’s (1977) work on “apprehension” would be more helpful if it took account of the large, important literature on anxiety (cf. Sarason [Ed.], 1980). Daly’s “apprehension” appears to be a sub-type, namely anxiety before the act, as indicated in the person’s own judgement.]

 2. 11 The second consequence is that “authorities on usage” disdain unbiased empirical discovery and rational argument in favor of bald assertions about “correctness.” Discussions of writing (e.g. handbooks on style and composition) freely mix general strategies, such as removing ambiguities and needless wordage, with the authors’ personal and social biases. Strunk and White (1979) weaken their reasonable advocacy of precision, clarity, and brevity (e.g., “keep related words together,” “omit needless words”) with a flurry of fussy distinctions (e.g., ‘compare to’ vs. ‘compare with’, ‘due to’ vs. ‘owing to’, ‘shall’ vs. ‘will’, ‘while’ vs. ‘although’) and petty dislikes (e.g., ‘anticipate’, ‘contact’, ‘currently’, ‘hopefully’). Whatever the authors don’t care for is stigmatized as “incorrect/wrong,” “awkward,” “loose,” “silly,” “absurd,” or “annoying”; writers who don’t agree are called “indiscriminate” and “illiterate” promulgators of “vulgarisms” and “abominations” (1979: 42-63). Similarly, other “authorities” — Richard Grant White, Jacques Barzun, Wilson Follett — denounce anyone who rejects staid prejudices and advocates language tolerance as a deranged anarchist scheming to bring doomsday upon us all (cf. survey in Finegan, 1980).. {13} The real goal of these polemics (aside from grabbing attention)¹ [Edwin Newman’s smug (1973) book simply uses literacy problems as a pretext to parade the travels and prominent acquaintances of its own author. Many of his anecdotes have little bearing on usage.] is to intimidate any counter-forces of language change or dialect proliferation that might endanger cultural elitism. The metaphors for adopting new usage are revealingly political (“sabotage,” “guerilla war”), legalistic (“infractions of rules”), social (“slumming,” “folksy”), moralist (“loose,” “perverse,” corruptions,” “sins”), and racist (“crossbreeding parts of speech”) (Follett, 1962: 74ff; Strunk & White, 1979: 45, 47, 52, 77, 81f). If public opinion agrees with the authority, it is praised as “logic,” “common sense,” or “established,” “modern,” and “preferred usage”; if not, it is damned as the popularity “ that will make language “as chaotic as a ball game with no foul lines” (Strunk & White, 1979: 44, 52, 57); or “a confusion of unchanneled, incalculable willawaws” (Follett, 1962: 75). The elitist undercurrent is usually concealed — after all, one must “not inject opinion into a piece of writing” (Strunk & White, 1979: 80) — but breaks forth when the general audience is abandoned as the final adjudicator. After decrying usage that makes readers “impatient or confused” and “wastes their time,” Strunk and White (1979: 78, 81, 85) abruptly decree that 

the writer [should] sympathize with the reader’s plight [...] but never seek to know his [sic] wants. The whole duty of a writer is to please and satisfy himself [sic],²[Mr. While is adamant that sexist language has ceased to be so, thanks to long use; he condemns as “boring,” “silly,” “nonsensical,” and “diffuse” both ‘he or she’ and ‘they’ as pronouns for indefinite persons; for him, anyway, the “writer” is “a man” (Strunk & White, 1979; 60f, 84).] and the true writer always plays to an audience of one.

 Yet most bad prose comes precisely from writers who rank their own mannerisms and convenience over the readers’ (cf. III.3.16; VI.1.25ff). 2.12 Such popular handbooks have had a wide impact.³ [The back-cover blurb on Strunk and White claims an stomach-turning sale of two million copies.] Their mode of dictating usage sets a bad precedent for society and the schools: accepting the welcome evidence from usage and public standards, and ignoring or vilifying the unwelcome. Correct usage is determined by decree and assertiveness. The average learner is rendered anxious and insecure by authoritarian attitudes, yet has no rational grounds for making decisions, because the promulgation of personal biases prevents a clear, workable consensus. Learning cannot be cumulative if every teacher legislates a new set of “correct” options (V1.3.2). Students must adapt to each new rule-set until they doubt the validity and rationality of any language judgements. Even beyond spelling, punctuation, and grammar, superstition holds that every language choice is either “right” or “wrong,” though the standards of rightness may  {14} never be clearly explained. To appear less vulnerable, many people react as defensively and violently as the “authorities” when their usage is questioned. Some composition students take all criticism of their writing as degrading and offensive, and desperately fight against change (cf. I.3.12, 15, I 9).

 2.13 This state of affairs is part of a pervasive educational attitude we can call the right/wrong dichotomy. Politicians and philosophers have traditionally advertised their views as the “truth” whose acceptance was more a matter of moral rectitude than of dispassionately weighing the evidence (I. 1.2). Dissenters are given the same vigorous flailing as the advocates of language tolerance and change (I.2.1 1). Science tends in the same direction to the extent that it ignores its own contexts (cf. I.1.2; 1.2.23.9; VI.3.21). Education has accordingly been set up as a relentless recitation of “right answers,” rather than as a forum for acquiring powerful, self-reliant strategies to devise, assess, and de-bug answers (Papert, 1980). If necessary, educational domains are fragmented, reduced, or distorted to maintain the dichotomy of right vs. wrong, and the simple, mechanical grading it enables. “Facts” are learned like mad, but not integrated into a meaningful framework (I.3.6). Literature, a domain that thrives on freedom of thought and expression (V1.1.7f), is studied as a endless drill on quotations, plots, or authors’ biographies. Composition, a domain for finding one’s own voice and taking stock of one’s knowledge (I.2.8.8; I.2.23.5), is disfigured as a self-conscious exercise in adhering to “rules”; monitoring the surface text becomes more important than delivering a purposeful message.

 2.14 A survey of several hundred freshman writers at my university turned up a veritable class menagerie of “rules” from previous English teachers (cf. VI.I. 12, 15). Some of the reported rules forbid the use of major options, such as opinions, adverbs, contractions, first and second person, the passive, the verb ‘to be’, quotations at the end of a paragraph, and questions in an argumentative essay. Other rules outlaw more specific options. No sentence can begin with ‘there’, ‘it’, ‘this’, ‘and’, ‘but’, ‘so’, ‘because’, or any conjunction or preposition.” No sentence can end with ‘it’, ‘is’, ‘to’, or a preposition.¹ [Like split infinitives, clause-final prepositions were outlawed because Latin, traditionally the language whose grammar served for other languages, doesn’t have them (cf. Postman & Weingartner, 1966: 47]. The blacklist of individual expressions included the transition words ‘however’, ‘too’, ‘in addition’, ‘furthermore’, ‘like’, and ‘because’; plus hedges and qualifiers, such as ‘perhaps’, ‘in my opinion’, ‘hopefully’, ‘anyway’, ‘a lot’, ‘a little’, ‘very’, and ‘really’. Still other “rules” prescribed numbers. A paragraph must have at least three sentences, and not more than five (cf. VI.I.15). No sentence may contain more than 30 words, and no paragraph more than 150 words. A high-school teacher in Indiana decreed that every research paper must have exactly 15 pages, {23}because that number was the average he had received in previous years.

 2.15 The origins of these rules are complex and diverse. Some rules reflect a helpless search for order and uniformity in a seeming chaos of choices and varieties. Some combat presumed markers of verbosity, obscurity, and pretense. Some are taken uncritically from old handbooks like Strunk and White (e.g. no opinions!) (I.2. 11).¹¹ Some are violent reactions banning language items that happen to get misused or overused — a response as simplistic and disastrous as the Biblical admonition to tear out an eye that offends you (Matthew 18:9; Mark 9:47). Some discriminate against conversational speech, or against particular dialects (I.2.7f; 1.2.23.4). Some simply make grading easier by sorting out the right from the wrong (I.3.14; VI.I. 15). Since these motivations are such a mixture of good intent, convenience, helplessness, and misinformation, the rules themselves are not likely to be consistently or rationally defended. Like the “authorities” on usage, teachers have the institutional leverage to legislate rules and to stigmatize people who deviate. Putting all injunctions in the same category of “good usage” creates a facade of unity and respectability to ward off challenges against any particular rule.

 2.16 The disadvantages of the rules must therefore be made clear. Objections can be raised on the grounds that the “rules” are:

 2.16.1 Incomplete. The rules add up to an oddly tattered patchwork whose gaps are greater than its fabric. For example, conjunctions are evaluated for some positions, but not for others.

 2.16.2 Inaccurate. The rules misrepresent what is demonstrably true of skilled writing, e.g. the dimensions of the paragraph. My students from Mimi were unable to discover in essay collections a single example that followed the formula their high school teachers had stringently required (I.2.14; VI. 1. 15).

 2.16.3 Authoritarian. Lacking a basis in observable usage, many rules merely embody personal biases. Bald assertiveness is considered just as respectable a basis for rules as serious research (cf. I.2.1lff; VI.3.25).

 2.16.4 Inconsistent. Some rules seem valid only for a single textbook or teacher: moving to a new class may mean unlearning them and learning a new set, lest the old ones lead to “errors” (I.2.12). Textbook authors may set down a rule and then ignore it in their own writing (even on the same page) (cf. VI.I.22). Learners conclude that the rules are arbitrary, and the compulsion to obey them unfairly distributed.

 2.16.5 Restrictive. The rule declares one option to be the only one, or misjudges a special case to be the general one. Motivations for individual contexts don’t come into view. The decision-making capacities for self-reliant use of varied writing options cannot be learned (1.2.12; cf. Youdelman, 1978).

2.16.6 Negative. The rules are mostly cast as “don’ts,” telling writers only what not to do. Such advice does not help writers decide what should be done, and evokes strong inhibitions. In consequence, students fear experimentation and creativity as traps leading to “errors.” No attempts are made to develop a personal voice (cf. 1.2.8.6). 2.16.7 Unworkable. Many rules are defined so vaguely that only a proficient writer could apply them. An example cited by Robert Gorrell (1965: 141) is “never include more than one thought in a sentence”; naive students have no criteria to identify thoughts or determine their borderlines. Sometimes, two rules collide and leave the writer no way out, such as simultaneous proscription of both first person and passive (cf. VI.I.22; Harris, 1979). 

2.16.8 Discriminatory. Many rules stigmatize specific features of minority dialects (cf. 1.2.7; 1.2.23.4). Justifications, if provided at all, usually conceal this striving behind some argument about logic or clarity. For instance, double negatives, a prominent feature of Black English (Scott, 1979), are attacked on the empirically false premise that they produce not an emphatic negative, but a positive (cf. IV2.55). 2.17 The effects of the rules can be considered next. First, accurate, comprehensive surveys of current usage (e.g. Quirk et al., 1972) are overlooked, and the regional, social, and personal variations they reveal are discounted. Second, by misrepresenting actual usage and undermining the credibility of composition programs, the rules impede learning more than they support it. Untrained writers misunderstand any advice as a strict rule (Rose, 1981). They get caught between conflicting rules (Harris, 1979). They invest great ingenuity and intelligence in the struggle to follow or devise arcane rules, and make errors that are ironically construed as signs of inability and unintelligence. (Shaughnessy, 1977; Bartholomae, 1980). And they may strain to follow rules until they get “writer’s block” (Rose, 1980). In such situations, trying out new skills seems too risky and frustrating to be worthwhile (1.2.23.3; 1.3.8, 22). The red marks on papers “are not attuned to the approximate ‘rightness’ of what the student has written, but rather to the absolute wrongness of the composition”; they “make no attempt to tell the student to what extent his writing begins to approximate acceptable English and rhetorical effectiveness” (Zoellner, 1969: 280). Writing loses all purpose, enjoyment, or personality, and turns into a tedious, mystifying ritual, game, or tightrope act. The tremendous concentration on minor “errors” obscures the students’ ability to implement major strategies. Eventually, students give up on literacy, like passengers abandoning a ship that sinks while the architect corrects the dining room clock on the Titanic.
   2.18 In many ways, literacy educators and composition teachers are caught in an acute dilemma (cf. 1.3.13; 111.3.30; IV.3.45; V.1.40; V2.3, 31; VI.I.9, 13; VI.3.2). {17} They teach effective, self-reliant writing in a society that expects them to embody and enact its diverse, inconsistent, and irrational language attitudes. Teachers may indeed be forced by parents or schoolboards to uphold odd standards of usage. Teachers who adopt an enlightened, tolerant stance on usage risk being denounced as irresponsible, dangerous saboteurs trampling on all standards (1.2.11) — intense diatribes are good for concealing indefensible, inaccurate views. Similar biases live on in textbooks, because publishers don’t want to offend those who decide what schools will buy, whether or not any expertise on language or writing influences the selection (VI.3.26).

        2.19 The solution to the dilemma will not be easy: extensive research to situate rules and errors in their proper social and psychological contexts (cf. I.3.8, 14). From a social perspective, we might probe public attitudes, self-images, writers’ anticipations of success or failure, prestige of dialects, and the literacy judgements discussed in 1.2.8. In their attempts to prevent language change and to uphold cultural supremacy, authorities on usage pick out far more “rules” and “errors” than are socially relevant, and may even disregard them in their own texts (cf. 1.2.16.4). Socially relevant, on the other hand, are usages that elicit judgements about a writer: obvious mistakes in spelling, punctuation, and grammar; unwittingly bizarre word choices; unintentional humor; ineptly pretentious verbiage; and so on. Students should learn about social consequences of usage, but not in composition courses that act as “laboratory sessions on linguistic and social conformity” (Larson, 1972: 635), or “maneuver students to internalize failure” (Kampf & Lauter, 1973: 23). Teachers should help students appreciate social pressures, but should not be arbitrary, unthinking enforcers of those pressures (V.2.31; VI.I.9, 13ff; VI.3.25). Teachers need to clarify public attitudes and work to make them more tolerant (Gere & Smith, 1979; VI. 1. 27).

 2.20 From a psychological perspective, we must explore the processes of producing and receiving texts to determine what advice would genuinely improve written communication. Procedurally, a RULE is a regularity to be observed unless the writer has a good motivation not to; and an ERROR is an occurrence which consumes processing resources unnecessarily by an unmotivated departure from such a regularity. The system of text production is so complicated and extensive that it is naturally prone to limitations and dysfunctions, such as processing overload  Errors are a normal side-product, not proof of low ability or intelligence. Writers merely need proper “debugging” techniques: methods for expecting errors and reacting strategically. Learning improves best by monitoring one’s own performance and applying the feedback (cf. III.3.2.2). Teacher-performed correction is not only unreasonably time-consuming, but also discourages students from learning to find and remedy errors on their own (V1.3.17).

 2.21 Research on rules and errors can be productive within the larger framework of an antecedent theory of communication and cognition {18} (I.1.12f). The standards of good writing can only be stated and defended in terms of the actions and interactions of writers and readers. Students need active, positive advice, not inhibiting prescriptions. If we shift our emphasis from the written artifact over to processing actions, we will come to understand the motives for our advice and evaluations (cf. VI.3). We can also clarify basic questions such as the order in which course materials are presented and tasks are carried out. For example, demanding a complete, well-written essay at the outset of a course violates the principles of learning psychology by taking “the most difficult part of the problem first” (Kintsch, 1977: 402). Success is more likely if students begin from their current abilities and edit their successive approximations into a finished product, stage by stage (cf. III. 1.26; V 3.40). Non-traditional students have a particular need for manageable sub-tasks (Roueche & Kirk, 1973: 68).

 2.22 Such research is all the more pressing at a time when both society and the schools are re-appraising their notions about literacy education. The “back-to-the-basics” movement, a signal of disenchantment with recent methods, can hardly help out unless we identify the truly basic skills: model-building, problem-solving, search, hypothesis-testing, resource distribution, and so forth (cf. Beaugrande, 1980a: 278f; VI.3.28). Rote memorization, repetitive practice, and authoritarian discipline are not “basic,” but simply ineffectual and alienating. Learning requires a cognitive reorganization of knowledge or skills to attain a more strategic design (cf. III. 1.28). Mere repetition is as likely as not to reinforce a non-strategic tactic and block off further learning altogether.

 2.23 The disappointing results of composition classes are occasionally explained by recourse to popular myths about educational potential (compare Combs, 1979). These myths are still strong enough that it is worthwhile to analyze the fallacies involved. For the sake of discussion, I will render each myth as a statement:

 2.23.1 “College students today are less intelligent and capable (or lazier) than in the past.” This myth is propagated in courses where traditional methods no longer work. Students who flunk composition often do well in such fields as computer science and chemistry whose complexity has greatly increased over the past. The problem is not lower potential, but a new diversity in background, expectations, and aspirations (I.2.2). For example, nontraditional students raise stronger demands that coursework be clearly relevant to a future career (Cross, 1974: 87). Early schooling (with its fixation on odd rules and mysterious errors) has led students to suppose that the effort expended on writing will not pay off, so that exertions are futile (cf. I.2.17).

 2.23.2 “College-age students have already completed their language acquisition before arriving in college, and major changes are no longer possible.” This myth was inspired by certain views on developmental learning. Schmidt (1975: 257) reports the view that “most of motor learning occurs in {19} children and that adult learning is merely a recombination of old habits.” Ausubel (1967: 307) reports the view that certain developments are possible only during “critical periods” and, if missed, must lead to “retardation.” Such claims reflect an unduly narrow and literal interpretation of findings on physiology, neurology, and maturation. Normal development is misunderstood not as the average (what most organisms do), but as the limits of the possible (the only thing that all organisms can do). Failure of an educational program is blamed on the hard-wiring of the human brain (cf. I.3.2; II.2.3ff). The best refutation is that many people do continue to learn language throughout their lifetime, and that developmental programs such as Mina Shaughnessy’s do bring about major changes in student performance.

 2.23.3 Still, cognitive and motor organization becomes steadily more complex throughout childhood, adolescence, and maturity, and thus harder to reorganize on any large scale, especially automatic actions (III.1.21). If some stage within this progressive organization should be mismanaged, or avoided as dangerous risk (I.2.16.6, 17; 1.3.8, 22), the later stages that presuppose it would be set up and learned in entrenched non-strategic modes (II.3.31). Repetitive practice would only make students fall further and further behind, and perhaps be barely literate at the end of high school (I.2.6). Properly designed training could preclude such a cycle by explicitly addressing the cognitive and motor reorganization as it progresses.

 2.23.4 Another source of this myth may be the attempts of some schools to stamp out student dialects. Dennis Baron (1975: 176) depicts the strange ideology of that crusade: “Those who would eradicate non-standard speech and writing operate under the delusion of a white man’s burden of absolute cultural supremacy. ‘Proper’ usage is seen as the outward sign of grace, while non-standard usage is regarded as evidence of physical and moral decay.” Though few teachers hold such extreme views today, some still misunderstand dialect features as indicators of mental limitations (I.2.7). Bereiter and Engelmann (1966: 39), not finding in the speech of Black children the categories of middle-class White English, argued that the children suffered from “a total lack of ability to use language as a device for acquiring and processing information.” Linguistic forms were used as evidence. “The speech of the severely deprived children seems to consist not of distinct words, as does the speech of middle-class children of the same age, but rather of whole phrases or sentences that function like giant words. That is to say, these ‘giant words’ cannot be taken apart by the child and recombined […]. The child’s pronunciation arises from his inability to deal with sentences as sequences of meaningful parts” (Bereiter & Engelmann, 1966: 34ff). If this view were accurate, then all speakers of Aymara, an Amerindian language centered in Peru and Bolivia, should also process no information. Aymara creates utterances as single long words, e.g. ‘Aruskipasipxañanakasakipunirakïspawa’, meaning, ‘I know it is desirable and necessary that all of us, including you also, keep communicating’ [data from Martha Hardman of the University of Florida]  Bereiter and Engelmann’s good intentions and sincerely helpful remediation programs were hampered by naively assuming the categories and differentiations of their own language and dialect to be privileged prerequisites of communication and cognition. The alternatives, e.g., using intonation and non-verbal signals instead of grammar, were not properly considered (cf. V.3.27).]

 2.23.5 “Composition is not a useful skill in today’s society.” The fallacy here is to overlook the importance of university training as a way to acquire cognitive processing strategies, not just a heap of facts. Writing offers the greatest opportunity for taking stock of and fitting together one’s knowledge (cf. Berthoff, 1972; Elbow, 1973; Lemke, 1974; Emigg 1977; Nystrand, 1977; Nystrand & Widerspiel, 1977; Olson, 1977a; Cooper & Matsuhashi, 1978; Murray, 1978; Irmscher, 1979b; Bereiter, 1980; Macrorie, 1980; Odell, 1980; Wason, 1980; Applebee, 1982; Shanklin, 1982; cf. 1.3.17; III.3.42; IV2.33; V.3.10, 15, 31). Literacy was therefore a major contributor to the evolution of social complexity and technology (I.1.13). Everyday speech is less adapted to assembling and reworking large-scale messages, or to preserving them for future consultation. Thus, most skilled professions in literate cultures demand writing abilities (Faigley, Miller, Meyer, & Witte, 1981).

 2.23.6 Writing might seem unimportant because university curricula neglect it. A national survey directed by Arthur Applebee (1981, 1982) learned that writing takes up 41 % of class time in English, and 48 % in the sciences, but mostly in short-answers, fill-in-the-blanks, or multiple-choice tests, where “right” and “wrong” answers are easy to grade (cf. I.2.13). Only 10% or less of class time is spent on composing extended texts, whose diverse, individual formats are harder to grade, though showing more reliably whether knowledge has been integrated and digested. 2.23.7 “Composing is an art, a special talent, and cannot be taught.” Steinberg (1980: 156) recalls “the romantic model”: “that truly creative writing is mysterious and non-rational, the result, ultimately, of inviting the muse rather than of employing consciously an ordered, rational process.” This myth probably stems from the origins of creativity, which begins by de-stabilizing everyday beliefs, questioning the obvious, or placing old bits of knowledge into new patterns (cf. I.3.10; III.2.15; Ghiselin, 1975: 357f). Until a new configuration of meaning falls into place, its inconsistencies resemble those typically found in irrational reasoning (III.3.15). But whereas irrationality preserves or perpetuates inconsistencies, creativity brings forth an insightful new order. Writing is therefore essentially orderly in its outcome, despite intermediary transitions in cognitive organization.

 2.23.8 “Non-traditional students lack the potential for academic success”. {21} This myth arose when non-traditional students found little help in “remedial” programs that merely rehearsed the tasks of traditional schooling all over again. But if their needs are met, non-traditional students can do as well as anyone else (cf. I.2.1). Educational research points up the importance of tailoring the school to the learner (cf. Hunt, 1972; Gordon, 1979; I2.21; II.2.32.4; V..8; VI.I.33ff; VI.2.26; VI.3.14), for example, allowing variable time for acquiring or performing a task (cf. Carroll, 1963; Bloom, 1974). Prior experience leads high-risk students to see achievement as a forced choice between total success vs. total failure (cf. Atkinson & Feather [Eds.], 1966; Holt, 1970). The SCOPE inquiry into student attitudes revealed that high-risk students typically have a need to expect success before they undertake a task (only a third of the conventional students felt this way). Erratic performance is natural in stressful areas, such as composition instruction. Still, I have found many non-traditional students bright, capable, and articulate, once they attain enough success to overcome their anxieties, inhibitions, and alienations (cf. Shaughnessy, 1977).

 2.23.9 Blaming failure on “low abilities” among learners is more convenient than developing more effectual educational approaches (cf. I.2.1; 1.3.2). Arguing from a simplistic model of “intelligence,¹ [Jensen’s (1973) model of intelligence has two “levels,” “I”and “II”: ability and memory for associative learning (I) vs. for conceptual learning, reasoning, and problem-solving (II). Level I is claimed necessary for level II, but not vice-versa — an assertion not supported by the evidence (Horn, 1976; cf. Grover, 1981). Jensen’s model sneakily legitimizes the independence and primacy of behavioral experimental psychology as over against cognitive psychology.] Arthur Jensen (1969, 1973) claims Black children to be genetically inferior. The inability of other researchers to equalize performance with less culturally biased tests was taken as further support of this outrageous claim (Jensen, 1980). What Jensen’s own fragmented, decontextualized “scientific” mentality prevents him from seeing is that high-pressure test-taking itself is a heavily acculturated middle-class activity (cf. Thorndike, 1968; Fishman, Deutsch, Kogan, North, & Whiteman, 1967; Roth, 1974; Marks, 1982). Tests “are unfair to culturally deprived children,” who, “in comparison with their middle-class agemates, have fewer test-taking skills, are less responsive to speed pressure, are less highly motivated in taking tests, have less rapport with the examiner, and are less familiar with the specific vocabulary and tasks that make up the content of the tests” (Ausubel, 1967: 313). Even if content items are shifted toward Black culture, the testing situation retains its fundamental imbalance. Moreover, the whole notion of “intelligence” is so poorly understood at present (I.3.2ff) that it is irresponsible to use test scores to slam the door on disadvantaged children.² [Case histories show that people’s lives and self-respect have been badly damaged by how they scored on the insensitive tests marketed by the Educational Testing Service, against whose judgements few students can effectively protest (cf. Nairn, 1980; 1.3.5f). The “verbal” parts of those tests are, in my judgment, especially arbitrary and questionable.] {22}

 2.23.10 “Compensatory education programs haven’t worked well in the past.” In early years, the value and direction of these programs was not clear (cf. Blocker, Plummer, & Richardson, 1965; Roueche, 1968). Direct evaluation was still rare (cf. Kendrick & Thomas, 1970). According to occasional probes, only a third of the non-traditional students entering special programs at community colleges completed the two years of study (Ludwig & Gold, 1969; Snyder & Blocker, 1970). But this discouraging record is misleading. Colleges and their staff were ill-prepared: “Remediation efforts […] consisted largely of watered-down versions of regular college courses. Each department in the college assumed responsibility for organizing and teaching the course, and in many instances instructors believed that the integrity of the department and the scholarly content of the discipline must be protected. Students, meanwhile, continued to fail and to resent even more the subject and the traditional manner in which the educational process was conducted. Instructors were assigned to remedial courses based on the pecking order of seniority and tenure in the department” (Roueche & Kirk, 1973: 11). The upshot was a paradox. The least experienced faculty had to teach the most difficult courses without appropriate textbooks, syllabuses, or compensation in lower teaching loads and higher pay. There was no terminology for explaining to these students what their problems were and how writing is organized. Traditional grammar is not an effective aid in such settings (cf. V.2.1, 16, 27).

 2.23.11 Under such conditions, compensatory education programs could hardly succeed. But when the programs are appropriately re-designed, clear benefits are attained, as Roueche and Kirk (1973: 82ff) have shown. Layout and materials must be genuinely innovative, simply repeating time-worn drills won’t help (Mellon, 1981: 56; cf. 1.2.22). Non-punitive grading alleviates the success-or-failure fixation mentioned in 1.2.23.8. Coursework should be self-paced to accommodate individual differences (cf. I.2.23.8). Course materials must be demonstrably relevant to the skills being acquired (I.2.23.1). Genuinely committed instructors should be assigned on a voluntary basis. Reliable counseling services should be provided by the college. And, finally, a smooth transition to regular coursework should be provided later on.

 2.23.12 “Soon, computers will take over all the writing from human beings anyway.” This myth reveals one among many popular misunderstandings of the nature of computers. The present volume, for example, was written on a “word processor,” in this case a primitive micro-computer with a text-editing software package; yet in no sense did the computer do the writing. It reduces the mechanical drudgery of writing and revising simply offers wonderful capabilities for drafting, formatting, word-searching, cross-referencing, indexing, and printing (cf. V1. 50). But it has no idea what to write about. {23} Full automatic text production programs so far write in a stilted, limited style below the versatility of high school writers, and function only as prototypes for research (e.g. Klein et al., 1973; Goldman, 1975; Wong, 1975; Meehan, 1976; Cohen, 1978; Davey, 1978; Mann & Moore, 1980). (However, unscrupulous internet firms do market ready-made essays students can buy — for at least $50 — and hand in as ‘their own’, hoping to hell the teacher hasn’t read it before.) Besides, even if text production could be computerized, humans would make a grave mistake in abandoning writing as a mode of consolidating and developing knowledge (I.2.23.5) and of finding an individual voice (I.2.8.6).

 2.24 These, then, are some obstinate myths about composition and its chances in the schools. The literacy crisis, whatever its origins and status may be, at least assures us that specious, convenient rationalizations of writing failure can no longer be accepted at face value. The proper recourse is rather to design realistic, workable models of the writing process, and from there, the coursework to guide that process for all learners, irrespective of their social and personal backgrounds.  

3. EVALUATION AND GRADING 

 3.1 Widespread anxieties about literacy are closely related to the heavy pressure in American education to quantify performance, as if learning would instantly cease without constant checking and measuring (cf. I.3.27). Throughout schooling, children endure batteries of diagnostics and tests that put a numerical value on everything conceivable. Test results are popularly misunderstood to be not a diagnostic for guiding a child’s evolution and expansion, but as a statement of the limits to which the child can aspire (I.3.4). The tests become self-fulfilling prophecies for an entire school career, and-when scores and grades are used in hiring beyond in professional life. Graduation from high school or college now depends on “competency tests,” especially in English (cf. Cooper, 1981b; Mellon, 1981; Myers, 1981; Odell, 1981). Frequently, such tests are devised by “educational theorists and researchers and curriculum specialists” who “slight or ignore the basic questions about competency testing in language, reading, writing, and media” (Cooper, 1981a: xiii). Hence, the tests reflect little awareness of current theories of cognition and communication (I.2.15; VI.3.2).

 3.2 Standardized testing has gone far beyond specific classroom tasks to become a widely respected measure of general intelligence and aptitude. A century ago, Francis Galton (1883) (a cousin of Charles Darwin) asserted that intelligence is an inherited, i.e., genetically determined, ability. This view was accepted by Alfred Binet, who devised an IQ test still in use today in its “Stanford revision,” and by a long succession of later psychologists (e.g. Spearman, 1904; Burt, 1940; Jensen, 1972). Marks (1982) maintains that the view survived because it was congenial to such ideologies as {23} Darwinian evolution and competitive commerce; but the view is still empirically unwarranted and theoretically confused. Even today, researchers have no secure consensus about the nature and constitution of intelligence (cf. the controversy in Sternberg & Detterman [Eds.], 1979). Some see it as a single general capacity, others as a set of more or less independent factors, and still others as the interaction of a general capacity with the sub-factors it supervises (survey in Das, Kirby, & Jarman, 1979; Sternberg, 1979). The “aptitudes” measured by the tests might well be chance products not based on any inherent personal merit or capacity, but rather on the conditions of the testing situation as such. The Federal Trade Commission found in a 1979 study that coaching schools for test-taking significantly improve one’s scores. But surely these schools don’t suddenly raise one’s intelligence!

 3.3 The psychological and social impact of the testing situation itself is precarious. Many tests deliberately pose tasks not found either in school or everyday life, as if abilities could exist in a vacuum, devoid of purpose and context (Cicourel, 1974). The tenet of experimental psychologists that “objective validity” is guaranteed when subjects don’t know the point of a test makes sense only if we assume that abilities are wired into the brain, waiting to be detected when people are caught off guard. Success in real life seldom depends on performing novel and meaningless tasks. Standardized testing eradicates the crucial distinction between what is known and what can be accessed and performed on a single occasion (cf. Kintsch, 1977: 442; 1.2.8.2). High-pressure performance under extreme time restrictions overloads processing and degrades performance (ef.III.1.17; III.3.3ff; VI.15). On such testing, people may underperform at least as much as they could overperform if they had the chance to practice ahead of time. Hence, the tests “measure what people cannot do, rather than what they can do” (Shuy, 1981a: 169). Whether or not overload occurs is related less to one’s intelligence than to one’s threshold of anxiety. Many famous thinkers do not work well under high pressure, yet average children have to do so if they want to reap the benefits allotted on the basis of test scores.

 3.4 Test-taking is a “social activity” (Roth, 1974) for which children are unequally prepared, regardless of their mental potential (1.2.23.9). Even if the scores did yield an accurate measure, they only gauge the status quo, not the child’s capacity for development (Fishman et al., 1967: 163). Yet neither the uncertainty about intelligence nor the irrelevance of the tests has prevented the misuse of scores to judge children and to stereotype their abilities at a fixed level. The child’s personal development is programmed to run a certain way, and, because the mind is influenced by the model it builds of itself, the child tends to conform to the level “predicted” by the tests. Standardized testing teaches the child to accept a mental profile obtained under unfavorable, uncontrollable conditions as a statement of immutable endowments, like the features of the face. If “unintelligent” children do something clever or remarkable, it must be an accident; “intelligent” children {24} doing exactly the same thing are revealing their “genetic” capacity. The reverse interpretation is placed on mistakes and failures.

 3.5 The psychological and social repercussions of testing intelligence and aptitude came quite unexpectedly. Standardized testing was originally conceived to be profoundly democratic, allowing capable students of all social classes to demonstrate their potential on equal grounds. But instead, a new caste system arose, based on test-taking skills; and not surprisingly, the system tallied excellently with the old one based on social status and wealth (Burt, 1943; Nairn, 1980). A College Board Report on the SAT scores of 647,031 students tested in 1973-74 revealed that scores corresponded perfectly to parental income. Aptitude and intelligence seem to be about as hereditary and predictable as money. Moreover, surveys show that aptitude tests, such as those administered by the Educational Testing Service (ETS), are poor predictors of achievement. A careful investigation of some 827 validity studies conducted by the ETS between 1964 and 1974 showed that “ETS aptitude tests on the average predict grades only 8 to 15% better than random prediction with a pair of dice” (Nairn, 1980: 59f).¹ [Specifically, the SAT is, on the average, II.9% accurate in predicting first-year college grades. According to the 1977 report for the Carnegie Council of Policy Studies in Higher Education, the accuracy of the LSAT is 13%, that of the ORE 11%, and that of the GMAT a mere 8% (Nairn, 1980: 60ff). ] Grades in turn have been found to be poor predictors of scientific or technical creativity in a later career (Taylor, Smith, & Ghiselin, 1975).

 3.6 Suppose now that intelligence is qualitative, not quantitative, i.e., to be described with attributes, not measured in numbers. Its key factor is the ability to treat any task by specifying high-powered procedures. As the child’s intellect evolves, these procedures emerge from a steady increase in the elaboration, differentiation, and complexity of cognitive organization (cf. Piaget, 1976, 1977). Intelligence therefore depends on whether the continual re-organizing of cognition happens to create the procedures for a strategically designed processing system — a development which educational institutions can hardly support if they understand it only vaguely. Usually, the child is left alone to make sense out of massive course content in math, social studies, history, etc., with whatever cognitive resources and strategies happen to develop on their own. It is assumed that force-feeding a child with facts and figures ensures an optimal mind. At every step, tests are administered, not to see if the child is thinking, but to see if the facts and figures were absorbed. Tests of “IQ” and “aptitude” complete the picture by gauging a few effects of a cognitive evolution barely addressed in its own right by school instruction. Obviously, this “intelligence” is more a by-product than a determiner of the child’s struggle to work in an environment of scrutiny and high pressure.

 3.7 In some areas, competition and comparison can be irrelevant, even {26} misleading. The abilities of two children might be simply different realizations of general human potential. The dimensions that make them comparable on a standard scale could be totally unrelated to their most important capacities. For example, creative achievement transcends the conditions in which it occurs, and succeeds precisely by setting its own standards (cf. I.2.23.7; Ghiselin, 1975). The insistence on standardized measures therefore necessarily bypasses creativity, and may even discourage it (cf. Getzels & Jackson, 1962).

 3.8 In real life, a significant accomplishment normally emerges from a painstaking series of approximations that are technically wrong or unsatisfactory, but point the way to something better (Papert, 1980). An initial discontent with the current state of affairs triggers a search for an objective whose nature, at the outset, may be wholly unknown. Progress is riddled with miscalculations, backtrackings, revisions, and rejections until the objective gradually comes into view. Similarly, entry into a new, unfamiliar domain of more complex skills is necessarily marked by an increase in errors over the previous stage of limited, but well-rehearsed skills (I.2.17; I.3.2, 24; III.1.23, 28; V2.15ff, 28; VI.3.13). However, conventional schooling sets its own rigid “learning objectives” above the child’s creative discovery, and is intolerant toward the provisional errors such discovery entails. As errors are constantly corrected, suppressed, and punished, children decide to content themselves with the old, familiar skills and to forego the unrewarded and stressful passage to new skills or creative insights (I.3.27).

 3.9 The assignment of grades in the classroom occurs in this larger context of testing and measuring. Here also, evaluation may rely on unrealistic, high-pressure situations where approximations are not tolerated and creativity is discouraged. Like test results, grades have the psychological role of interpretive frameworks that both teachers and learners apply to daily activities (I.3.4). The accumulating series of grades singles out numerous learners, perhaps permanently, as “average” or “below average,” i.e. as “low-ability” persons to whom education typically offers little respect or resources (cf. I.2.23.8ff). The Darwinian “grade curve” aggravates matters by forcing superior ratings to be balanced out with inferior ones. Even the “C” grade for the large group in the middle, though designated “average,” is widely considered stigmatizing (V1.3.5).

 3.10 Grading is another manifestation of our society’s conviction that competition should be the measure and goal of all things. This credo has brought great advancements in business and sports, yet is hard to apply to intellectual or creative faculties that transcend normal standards (I.3.7). In some areas, competition may level cognitive originality and individuality simply by failing to recognize and reward it; ratings are based on aspects that make each learner’s achievements comparable to everybody else’s. In sports, players do best within fixed rules and scoring techniques. Cognitive growth flourishes when a learner decides to break old habits of thinking and find new measures for new achievements. {27}

 3.11 Grading also shares with testing a predominantly negative outlook: responding not to the total achievement, but to the number of superficial errors that can be detected and counted. Usually, the mere quantity of errors, rather than the processing slips or overloads they reflect, decide what the grade should be. Yet studies of cognition show that errors are a normal, inevitable by-product of efficient processing (cf. Norman & Shallice, 1980; Norman, 1981; Neisser, 1982). The expectations and assumptions built into rapid, automatic actions are perforce wrong some of the time. Error quotas are marginally relevant as long as debugging skills are available (Papert, 1980). Thus, negative grading is a form of punishment for being human. Learning progresses much faster if grading reflects the students’ accomplishment, not the number of mistakes. Achievement cannot be meaningfully defined or pursued as the mere absence of errors which, as I stressed already (I.3.8), are necessary by-products of significant advancements and new skills.

 3.12 The problems of testing and grading are extremely acute in the composition class. Here, the artifact to be graded is a personal processing outcome whose real complexity is utterly belied by a grade on a scale with anywhere from five to thirteen slots (depending on whether the record office allows “plus and minus grades”). Teachers may append commentary, but students have become conditioned to notice and react to the grade itself. The personal nature of writing reinforces the psychological repercussions of red marks and bad grades, as if, once more, the individual learner were on trial (cf. I.2.12).

 3.13 The composition teacher is again entrapped in a dilemma by popular attitudes (cf. I.2.18). Parents, school boards, administrators, and students clamor for the simplistic numerical measurements grades embody. Protests that literacy is not quantifiable in such a way go unheard in the scramble for the magical “grade point average” that stands (or once did) at the threshold of high-paying professions. Understandably, composition teachers are forced to navigate within this narrow-minded outlook. By attending only to the paper as an artifact, and overlooking the processes that produced it, teachers have at least some evidence to justify their ratings. Teachers may be well aware that there is no such thing as an “A paper” or an “F paper” independently of the context of its production, and yet invoke this fiction as protection when called to account. Just as no text is meaningful apart from context, no text has value without context. Like the scientists depicted in I.1.3f, teachers have to supply the context in order to reach any evaluation at all. The only question is whether this context is expounded well enough that students can also supply it and judge by themselves the merits of their writing. In my view, this additional step is the most essential precondition to resolving disputes and uncertainties regarding literacy inside and outside the schools (V1.3.17).

 3.14 All too often, grading was based on a counting up “errors”,” i.e. violations of such “rules” as were surveyed in 1.2.14ff. These error sums {28} were placed on a “curve,” and the grades were assigned to yield the proportions which rendered the curve symmetrical. However, this method would be objective only if the notion of “right vs. wrong” were at all appropriate for the majority of events in text production. In reality, composition teachers are hard pressed even to agree on common mechanical errors (cf. I.2.lif, 16ff; 1.3.17; V2.30, 32, 39; VI.I.12). In Greenbaum and Taylor’s (1981) study, teachers showed little consensus in recognizing and naming errors: the judgements of most teachers were in the minority on at least some examples, and over 35% of the labels given were inaccurate. Moreover, the tendency to find non-existent errors proved especially strong, as Deaux (1981: 76) found in another project, among inexperienced instructors. Paper-correcting expends atypically heavy resources on scanning the “shallow” text levels (in the sense of III.1.13) for events that would be perfectly acceptable under more natural conditions. Greenbauni and Taylor (1981: 173) cite the “anxiety” of failing “to catch all the errors,” intensified by “insecurity about grammar and usage” (cf. I.2.10). Hairston (1981) found an even greater diversity of attitudes about errors among professionals in business.

 3.15 Consequently, most composition specialists concur that grading should not be determined by counting errors. They recognize that a negative, reactive orientation is not effectual in teaching (cf. Porter, 1962: 15; Zoellner, 1969: 280; Irmscher, 1979a: 148ff; Kinneavy, 1981: 67). Mechanical errors are still commonly marked, except in special programs (e.g. Staton, 1981), though many teachers assign no great importance to the markings (Applebee, 1981: 74). However, students apparently need some retraining before they cease to be still intensely preoccupied with the error as a signal of intellectual or moral breakdown. They focus at once on any visible marks, sometimes ignoring the larger issues of evaluating and revising prose.

 3.16 According to Applebee’s (1982: 372) survey, 58% of the teachers comment on content organization. But a “content” grade not based on mechanical errors is no simple issue. It may be partly a rating of the ideology inferred from the paper, i.e., the conceptual framework in which the process of “ideation” (in the sense of III.2.14) takes place. This ideology emerges from one’s “belief system” (cf. III.3.15) about what is true and valuable in the world (that is, in one’s model of the world, I.1.8). This standard is precarious when the students do not share the teacher’s ideology and come to “the wrong conclusion” (Fishman et al., 1967: 161). Or, they may try to represent an ideology which is not theirs, but which they have come to know through long experience with teachers (cf. III.3.30). It is ominous that speech in behalf of views one doesn’t believe in has a noticeably higher percentage of errors (Mehrabian, 1971); students may well underperform if obliged to write insincere essays.

 3.17 The “content” grade may also reflect{29} the writer’s conceptualisation strategies, i.e., ways for organizing knowledge (cf. III.2.20ff; VI.2.27ff). These strategies figure prominently in recent rhetorically-based approaches to composition (e.g. Christensen, 1963; Winterowd, 1970; D’Angelo, 1979a), looking back to a rich tradition in argumentative rhetoric (cf. Perelman & Olbrechts-Tyteca, 1969; Corbett, 1971; Kinneavy, 1980). Teachers want good writing to respect “logical standards”: induction, deduction, consistency, and procedures for drawing conclusions. Jane Harper Yarbrough found that teachers in high schools and at the University of Florida labeled content problems in selected passages from student writing as “faulty logic” more often than in any other terms. However, the label was used for passages whose problems bore little resemblance to each other. The teachers apparently treated “faulty logic” as a cover term for all manner of problems in reasoning.

 3.18 Plainly, a “content grade” has merit only if based on clearly defined aspects. Yet “except for occasional brainstorming sessions reported by about a fifth of the teachers, there is no systematic attempt to provide students with strategies for recognizing what information might be relevant” (Applebee, 1982: 373; cf. VI.2.32). Rhetorical approaches could consider the use of hierarchies: co-ordination, subordination, addition, contrast, etc. (cf. Christensen, 1963, 1965; Pitkin, 1969, 1977; Grady, 1971; Winterowd, 1970a; D’Angelo, 1974; Nold & Davis, 1980; VI.2.22); or the performance of discourse actions: stating, restating, expanding, qualifying, conceding, evaluating, and so on (Larson, 1967). Logical approaches could consider the accuracy of facts — an aspect teachers favor (Applebee, 198 1: 74). However, stressing facts is inappropriate for some types of texts and situations, e.g. the personal essay, the journal, or the creative story; and may be unduly colored by issues of ideology (I.3.16). Literacy shouldn’t be uncritically equated with knowledgeability (cf. I.2.8.8).

 3.19 That teachers mark an error or a problem, either in mechanics or content, is itself a socially significant action (cf. Kroll & Schafer, 1978; Bartholomae, 1980; Butler, 1980; Williams, 1981 a). They must first respond to it as an error in their social role as a higher authority with legitimate and relevant expertise. They must also decide the consequences: grades, advisement, revision, etc. In an authoritarian setting, rules, errors, and punitive grading (cf. I.2.16) are emphasized at the expense of the students’ insights or recourses. However, the complexity and variety of language usage renders such a stance precarious. English teachers are atypically inclined to respond to errors, whence their tendency to mark non-existent ones (I.3.14). Students are subjected to unnecessary, intimidating language intervention normally reserved for young children. This experience is unsettling for adolescents who feel stigmatized as childish. The tendency to be defensive and make excuses (I.2.12) is presumably a reflex to protect the students’ emergent self-identity.

 3.20 The authoritarian approach is also undercut by the lack of consensus {30} among authorities and evaluators about “rules” and grading standards (cf. I.2.16.4). Diederich, French, & Carleton (1961) had freshman papers rated on a scale from one to nine by teachers in composition and in social and natural sciences, plus writers, editors, lawyers, and business executives. The results fluctuated bewilderingly. 30% of the papers received every grade on the scale, and 100% received at least five different grades. The contention that no paper has value without context (I.3.13) was strongly confirmed.

 3.21 Since the authoritarian approach is rationally untenable, the alternative must be a relationship of negotiation and equality. Here, the instructors’ advisement and evaluation strives to represent a socially general response by expounding and reacting to the standards of professional writing. Fair grading systems can be established on imperatives like these:

 3.21.1 Task-relevance. Consider the importance of instructional goals and the relative difficulty of their attainment.

 3.21.2 Achievement: Gauge the effort expended and the results attained by each student.

 3.21.3 Accessibility: Select and grade tasks the typical (not just the “talented”) student can realistically be asked to encompass.

 3.21.4 Transparency. Define all criteria such that the students as well as the teacher can apply them reliably.

 3.21.5 Input-output: Reward only the skills that are being actively taught.

 3.22 Traditionally, teachers have rated a paper by comparing it to: (a) an idealized version; (b) the version they would have written; (c) prestigious texts (e.g. literature, essays, science reports, etc.); or (d) other student papers obtained on the same assignment. These comparisons are oriented toward versions many students couldn’t possibly write: this remoteness of the frames of reference violates the imperatives of accessibility and transparency. To demand that students have the cultural and social prerequisites for such comparisons violates the imperatives of achievement and input-output. To assume that all students expend the same effort for a given result violates the imperative of task-relevance. Like the marking of errors, such comparisons rate the paper for what it is not (cf. I.2.16.6), rather than taking it seriously as a stepping-stone to progress. A creative paper shouldn’t be punished just because the unskilled writer doesn’t reach the ideal. Moving to a higher plane of writing skills naturally leads to a temporary rise in the errors and problems until full control is attained (Bereiter, 1980: 76; Elbow, 1981: 100f; Keech, 1981: 59f; cf. I.3.8).

 3.23 Fair grading should measure the individual student’s progress at that particular stage of development (imperatives of task-relevance, achievement, and accessibility). Moving from incompetent to average means confronting a vastly more complex and unmanageable problem domain than moving from fairly good to very good. It seems unjust to rate the {31} former progress with a “C” and the latter with an “A” solely because of an abstract quality difference in papers produced. Instead, we can base the grade on the difference between samples produced at the start of the course vs. those produced at the end. This policy requires techniques that use performance to infer processes and problems. The evidence obtained should not be treated as a judgement of the writer’s intelligence or abilities, but only of his or her current skill level. A single strategy, e.g. relying on the habits of spontaneous speaking (cf. V.3.10ff), might produce diverse error types, whereas several strategies might produce just one error type, e.g. spelling mistakes (V.1.14ff, 40ff). Errors or problems so diagnosed should not just be counted up for a grade (I.3.14), but should be used to guide individual advisement and training modules (imperatives of accessibility and input-output). Only errors or problems that persist throughout the course despite explicit, workable training should count toward the final grade. Total failure would scarcely occur unless a student refused to participate (I.2.5; 1.3.27).

 3.24 Surely no grade can be fair if students have no realistic chance to control the situation. Therefore, students must learn to judge and revise their own work. They should expect errors and problems as natural side-effects of entering an unfamiliar domain of skills (I.3.8, 22). Practice in editing helps to break down the large task of writing into manageable subtasks (imperatives of accessibility, transparency, and input-output)-particularly useful for non-traditional students (I.2.21; VI.3.13f). Individualized instruction allows the flexibility to deal with the particular skills and problems of students from varied backgrounds. We can construct and update a personal profile for each learner, so that needed training is selected and placed in focus. In my experience, a modest, tolerant, and readable workbook is helpful for showing how to revise. Individualization also enables personal negotiation of topic (cf. III.2.13). A “pointless” topic invites a “meaningless paper” (Booth, 1975: 75), ¹ [11 Applebee’s (1982: 374) national inquiry turned up “impossible topics” in English classes, such as this one for ninth-graders: “in a well-organized essay of 200-500 words, answer the question: how is Odysseus a model for youth of all times?”]/ since boredom reduces the total allotment of cognitive resources that writers are willing or able to make (cf. I.3.27; III.3.2.9). Motivating topics make the whole task easier and more fluent (cf. Graves, 1973; Keeney, 1975), and decreases the processing overload unskilled writers often encounter (cf. I.3.3). Though professional writers in real life may not be free to choose their own topic (Redish, 1981: 155), they are seldom asked, I think, to write on topics they have never studied or thought about.

 3.2.5 Unskilled writers learn more from revising and reworking an unsuccessful paper than from throwing it aside with a feeling of incompetence and failure. By learning to edit, the students realize that errors need not be {32} stigmatizing or stressful. Real editing (rather than mere recopying) requires concrete editing tactics for specific problems. I can then indicate which problems a paper contains, but leave the students to detect and resolve them. This use of modules supports selective focus, bringing one sub-task into the center of attention, and setting others temporarily aside. Students work more confidently and effectively if they don’t try to watch for too many things. In class, students use other people’s samples in a handbook (Beaugrande, 1982a), so that no personal attacks are implied. (Even anonymous passages from the current class can elicit resentment.) Students are advised about their own work in an interview setting, where they are less conspicuous and hence less sensitive. I have found that untrained writers gain crucial self-confidence from the tangible success attained via revising, and lower their anxieties about writing. Presumably, these changes in outlook reduce the tendency to perform below one’s knowledge and abilities (cf. I.2.8.2; 1.2.10; 1.3.3), and account for the observed disappearance of mechanical errors without remediation.

 3.26 In-class quizzes consist of short passages to be revised. Ratings are given only for the factors currently in focus. Other problems or mistakes can be marked, but not rated. If a quiz involves several factors, students are advised to edit in successive stages. This training is continually transferred over to their own writing. I withhold definitive grades until the student decides that a version is the final product. Revision is encouraged as long as a paper indicates weaknesses that could justify a low grade. This method tries to live up to the grading imperatives which were advocated in 1.3.21. The revision modules make it plain what goals are considered important and difficult (task-relevance). The use of personal profiles records individual progress during the course (achievement). Explicit revision tasks are readily understandable and manageable for the average untrained writer (accessibility and transparency). No prior assumptions are made other than that students are speakers of colloquial English.²¹ Finally, the end result is a set of skills that were explicitly addressed and practised (input-output).

 3.27 Many school settings are influenced by the facile assumption that students are naturally lazy (cf. I.2.23. 1). Incessant tests and quizzes are rationalized on the grounds that students won’t work unless you threaten and scrutinize them constantly. This myth is “a consequence of attempts to force an irrelevant curriculum on students” (Combs, 1979: 73ff):  

Seeing no meaning in what they were asked to learn, students were not motivated to learn […] it took a lot of effort to work under such conditions. […] Teachers, on the other hand, confronted with unwilling learners, found it necessary to increase the pressure to make students learn. Kids will loran difficult things without the use of coercion {33} if what must be learned is (1) relevant to the present needs of the student, (2) paced to current status and capacity, and (3) interesting and rewarding to the learner.

 Composition will be very hard if it is reduced to a stressful, frustrating confrontation with odd “rules” and “errors” (cf. I.2.17). The students who need the most help are the most endangered by any attempts to try out new words, phrasings, or styles. Students can be genuinely overloaded by writing tasks devoid of meaning and purpose, and will understandably try to escape them. My own experiences with “problem” learners has invariably been that they were not lazy or stupid, but alienated and unmotivated. If I gave them tasks of clear relevance, they performed perfectly well.

 3.28 Admittedly, designing new, individualized instruction increases the load of the teacher as it reduces that of the learner (cf. V2.27; VI.3.5) The additional work can be offset in several ways. (a) a drastic reduction in teacher-performed correction of papers in favor of student-performed editing; (b) strict and well-justified limits on the number of students an instructor supervises; (c) a shift away from writing many different papers toward revising until superior products emerge; and (d) a reduction in class meetings in exchange for interview time. Classes can be devoted to issues that concern the majority of students. So far, the approach has brought about encouraging improvement among the large middle group whose skills would conventionally be classified as “average” (with a hint of something gone wrong, cf. VI.3.5). Though entering the course in a mood of inadequacy, anxiety, and pessimism, they are surprised to find writing manageable after all. Their attitudinal and cognitive change is frequently impressive, particularly in comparison to students who either refuse to participate in the training, or who are proficient already and don’t genuinely need such a course. These two groups are comparatively small in my classes.

 3.29 Learning to write, and indeed any act of writing, is a process with no point of definitive completion (cf. III.1.26; VI.I.25; VI.3.28). In the same way, we cannot set a threshold where a composition coursework is indisputably finished, any more than a threshold where a paper is perfect for once and for all. High grades should reward significant progress, though a teacher’s ultimate ideal may not have been reached (cf. I.3.21). The proper conclusion to instruction is the point where the students grasp the writing process well enough to direct their own future evolution. The best teacher is the one who becomes expendable because the students have taken over his or her role for themselves.  

4. TEXT AND DISCOURSE AS EVENTS 

 4.1 The previous sections addressed the context of a scientific inquiry into text production and writing. The objectivity, consistency, and generality to which science aspires are to be found in processes, not artifacts (I. 1). {43} Such a science has a responsibility toward society: creating a corpus of knowledge about language to replace public attitudes detrimental to literacy education (I.2). This corpus in turn supports a justifiable and equitable methodology of evaluation and training (I.3). In this context, composition can profit little from a sheltered preserve of theory without practice (V1.3). We must rather account for the real human activities of producing and receiving texts in the social and psychological frameworks of normal life Consequently, we must begin by establishing that texts and discourses are events, not the static objects presupposed at the outset of the inquiry.

 4.2 Language is a SYSTEM — an assembly of elements whose FUNCTIONS are their contributions to the workings of the whole. Language in use is a dynamic system moving from one STATE to another. The language system is composed of LEVELS: concurrent modes of organization for selecting and combining communicative elements (sounds, words, syntactic patterns, concepts, ideas, etc.) in the various phases of processing (cf. III.2.5-40). These levels operate as sub-systems whose interaction constitutes the action of the main system (III.2.3ff); mapping comprises the processes that correlate the elements and functions of one sub-system with those of another.

 4.3 Linguistics studies language with the help of taxonomies: sorting schemes that list each element according to its identifying features (II.1.5). This approach works best if the system is compact and the features are plain; otherwise, headway is slow and schemes are periodically replaced. Taxonomies of the sound system in phonology have proven more durable than those for syntax, semantics, and pragmatics (II.1.6) — domains whose “features” are hard to define (III.1.16). The real events of language are not just abstract clusters of static features, but the outcome of human processing actions. We should not stress theoretical analysis to the exclusion of active synthesis (I.4.14).

 4.4 Here, systems theory helps to situate language in the larger context of communication and cognition. Theories encompassing that context are properly antecedent to theories of language (cf. I.1.12). Linguistic syntax is a special case of linear processing: a mode of behavior and a mode of intelligence (cf. IV.1.4). Semantics is a special case of the processes of conceptual processing: creating, storing, recovering and utilizing meaningful knowledge (cf. III.2.20ff). Pragmatics is a special case of goal-planning: setting up an intended state of the world and implementing steps to attain it (cf. III.2.6ff). These definitions stress that language is one mode of processing among human actions at large (cf. II.2.32. I ff).

 4.5 A system in use is being actualized, i.e., some of its options are being activated (taken from storage and implemented). Any resulting artifact is itself an actual system, as opposed to the virtual system of options awaiting use. Text production is one example of actualization, and the text {35} itself is one example of an actual system. Any further utilization of a text, such as reading, interpreting, quoting, etc., is also actualization and activation, though under new conditions affected by the original event of production, and by the surface text as artifact (if available) an illustration of intertextuality (I.4.11.6; cf. II.2.22; III.1.1.9). Actualization includes selecting options and combining them into STRUCTURES, where a “structure” is two or more elements with a relation obtaining between them (cf. II.1. 1; IV 1. 1). (Some authors prefer the term “structuration” over “actualization,” e.g. Wienold, 1971). A structure treated by processing as a single unit can be termed a CHUNK, and may range from a LOCAL (small-scale) MICRO-STRUCTURE, over to a GLOBAL (large-scale) MACRO-STRUCTURE (cf. III.1.10). A chunked micro-structure constitutes a MICRO-STATE of the actual system, while a chunked macro-structure constitutes a MACRO-STATE.

  4.6 The distinction between virtual and actual systems is crucial for realistic theories of cognition and communication, because the nature of any system is determined by the uses people make of it. Language items and configurations are INSTRUCTIONS about what to do during processing (cf. Weinrich, 1976). Alongside the resources of the system, users need PROCEDURES to access them, where “procedure” is defined as a configuration of steps carried out in order to obtain a result. For instance, the GRAMMAR of the language only defines itself when put to use (Lenneberg, 1975; I.4.10, 14; II.3.14; V.2.17, 27). The LEXICON is, in a virtual perspective, the repertory of words and expressions available in a language, and could be envisioned as an ordered listing like an everyday dictionary with an alphabetical format, plus implicit and explicit cross-references (implicit ones appearing when one word is defined with other words). The repertory can be further classified into CONTENT WORDS (nouns, verbs, adverbs, and adjectives) with a fairly stable conceptual content, and FUNCTION WORDS (prepositions, conjunctions, articles) whose contribution is largely to provide a framework for the content words (cf. Miller et al., 1958; Bolinger, 1975).¹ [Even the classic dichotomy of content words vs. function words is fuzzy in some areas. For example, pronouns are usually (but not always) derivative upon noun phrases for their content, yet assume the same grammatical roles as the latter. I suggest a differentiation in terms of memory search grids in III.3.13.] In an actual perspective, we must explore what procedures allow entries from the lexicon to be accessed, so that words can be learned, stored, recognized, and retrieved (cf. 111.3.10ff; VI.19, 28). Here, the meaning of a word is not a neat, stable definition, but the result of a cognitive operation. Knowing what a word means is thus having an ordered set of hypotheses about processing actions to perform on the word. The ordering of these hypotheses is strongly influenced by current context, not just by the sequence of definitions offered in a dictionary. {36}

  4.7 These considerations have led to recent discussions on classifying knowledge into DECLARATIVE (i.e., statements) vs. PROCEDURAL (i.e., operations) (cf. Winograd, 1975; Bobrow & Winograd, 1977; Goldstein & Papert, 1977; Winston, 1977). However, it would hardly be economical for humans to maintain two separate stores. More likely, they maintain one unified knowledge store so richly organized and interconnected that it can be accessed and used both as substance and as actions. Knowing what something is relates closely to knowing how to do things with it.¹ [Compare Ryle’s (1949) “knowing how” vs. “knowing that”; and Shiffrin and Schncider’s (1977) “actional” vs. “informational” knowledge.] One of these perspectives may dominate the other on a given occasion, depending on the exigencies of processing (cf. IV2.33). Active processing MEDIATES between one’s knowledge and the current situation (cf. I.1.7; 1.4.11.5f; II.2.10, 20ff; III.3.2.10) and decides what is RELEVANT at that moment (cf. III.2.15, 20; III.3.14; VI.2.2f, 6f, 11, 32).

  4.8 If communication does not consist of passing stable word-meanings back and forth, how do people know what they mean? The answer can only be that processing itself, far from being chaotic or random (I.1.3), must have a sufficiently well-structured, complex order to manage so many texts and communicative situations. This order guarantees that different people sharing a pervasive cultural consensus normally attain a comparable processing outcome from the text or situation at hand (cf. I.1.7). Processes are strategically organized to watch selectively for relevant indicators and to fit powerful procedures to current specifications (cf. Bobrow & Winograd, 1977). These procedures could be either ROUTINES (habitual actions for everyday contexts) or SPECIALISTS (actions that are triggered only when specific conditions are met) (cf. III.1.14, 27; Winograd, 1972; Minsky, 1979). A skilled text producer can activate a “logic specialist” that checks for consistency (II.3.3); a “scheduling specialist” that monitors the order in which things are done (III.1.23; III.3.4.9); a “style specialist” that controls word selections (cf. III.2.1 1, VI. 1.6); and so on. The specialists would remain inactive until context requires them. In absence of special circumstances, processing can apply a DEFAULT (an item or operation selected unless it is ruled out) or a PREFERENCE (a choice normally made over others). Defaults and preferences can be overridden as seems expedient. People understand each other to the extent that their processing is relatively uniform in its preconditions and consequences.

 4.9 The TEXT can be defined as a naturally occurring manifestation of language, i.e., as a communicative language event in a context. The SURFACE TEXT is the set of expressions actually used; these expressions make some knowledge EXPLICIT, while other knowledge remains IMPLICIT, though still applied during processing. A DISCOURSE is a set of texts considered to be mutually relevant. The totality that bears on a discourse can be {37} called the UNIVERSE OF DISCOURSE. The text and the discourse are by nature SYSTEMIC, i.e. work as a system (I.4.5). As a text is produced or received, the system is actualized and traverses a series of STATES. Under the conditions outlined in 1.4.6-8, the organization of any state depends partly on the virtual repertory of elements and options, and partly on the current conditions. In terms of cybernetics, the system regulates itself as its states succeed one another, either by “assimilating” an event into the system, or by “accommodating” the system itself (cf. Piaget, 1977). Any marked change elicits a compensatory change somewhere else in the system. Thus, special demands of context can be met with a suitable adjustment. For example, a word used in an idiosyncratic way, e.g. ‘sisters’ in Shakespeare’s line

  (1) Her art sisters the natural roses (Pericles, V, i, 7)

 is understood according to its setting. Participants in communication construct and test hypotheses based upon their expectations about what normally happens. In effect, people constantly build models of what is going on (I.1.14), and fit in whatever evidence they can gather. The models are only approximations (I.1.10), but they enable communication to proceed without having to spell out every bit of context.

  4.10 Modern linguistics has been most concerned with virtual systems (II. 1.2). For instance, the abstract “grammars” of the 1960’s can be viewed as statements of the syntactic expectations shared by the speakers of a language. Such a grammar could be consulted to decide if a sentence is “grammatical” (II.3.5, 14). However, such judgements are abstract in the sense that people (a) are most certain of their grammar when they actively use it; (b) need not agree about the ultimate borders of the grammar; and (c) given proper motivations (social roles, cognitive overload, inattention, stress, etc.), can produce sentences they or others would not consider grammatical. Since the text is defined as a real event, its “grammaticality” affects its status less than do the standards based on its relevance to the communicative situation. EFFICIENCY is the utilization of a text (by producer or receiver) with the least outlay of effort (I.2.8.3). EFFECTIVENESS is the degree to which the text has an impact on the situation and so forwards one’s chances of reaching one’s goals (I.2.8.4). APPROPRIATENESS is the degree to which one’s choices fit the current setting. These standards may fluctuate together or independently. Effectiveness might call for strange, attention-getting devices that arouse interest, but decrease efficiency. The text producer should strive for the level of difficulty appropriate to the audience.

 4.11 The notion of TEXTUALITY serves better than “grammaticality” as the criterion that distinguishes between a text and a non-text. Textuality can be defined according to at least seven standards (cf. Beaugrande, 1980a; Beaugrande & Dressier, 1981; and references there): {38}

 4.11.1 COHESION subsumes the procedures whereby the SURFACE TEXT is organized as a sequentially related configuration of language items. Cohesion rests upon syntactic formatting (reconstructable via “linguistic rules”) and upon such procedures as recurrence (re-use of the same expressions); paraphrase (presenting the same content in different expressions); parallelism (filling the same surface patterns with different items); pro-forms (place-holders such as pronouns); ellipsis (an incomplete pattern whose omitted items are readily supplied from the setting); and junction (connecting words to link sequences) (cf. IV.2.37-44, 51f).

 4.11.2 COHERENCE subsumes the procedures whereby a configuration of concepts is assembled into a TEXTUAL WORLD: the total knowledge activated while processing the text. A CONCEPT is a configuration of knowledge characteristically stored and activated together. PRIMARY CONCEPTS are the most basic to the organization of meaning (actions, events, objects, and situations); and SECONDARY CONCEPTS serve to differentiate and delimit the primary ones (time, location, cause, reason, instrument, and so on).¹ [An empirical typology of concepts or propositions has only begun to emerge from research (cf. C. Frederiksen, 1975, 1977; Beaugrande, 1980a). Any typology is likely to have drawbacks because concepts can be constituted in context, not simply drawn from an immutable list (cf. III.3.14).] The PROPOSITION can be defined as two concepts plus a relation obtaining between them (I.e., a conceptual structure); the RELATION determines how the concepts are mutually relevant, e.g. that one “event” is the “cause of” another “event.” Several propositions become a conceptual CHUNK when they are combined and processed as a single unit (I.4.5). REFERENCE is the relationship of a concept or chunk to a real or imaginable entity in one’s model of the world (cf. II.2.16; III.2.22). In principle, reference is an aspect of the whole text-world model, not of single words. This model includes commonsense world-knowledge not made EXPLICIT in the text, but contributed by various processes: SPREADING ACTIVATION, where an activated concept spreads its status to whatever is closely related in memory storage; INFERENCING that builds bridges among explicitly activated concepts (e.g. reasoning about causality); and by UPDATING that changes what is true in the textual world as knowledge is presented. These processes are supported by preorganized patterns of world-knowledge in memory: a FRAME is an array of knowledge (e.g., what one knows about houses); a SCHEMA is a progression (e.g., what happens if you build a house or walk though it from room to room). All these patterns and processes make it clear that coherence must not be confused with cohesion.² [Some researchers conflate cohesion with coherence perhaps because predicate logic obscures the distinction (cf. critique in Morgan & Sellner, 1980). For example, Halliday and Hasan’s (1976) “lexical cohesion” is a misleading term to the extent that “cohesion” is in the text, not in the “lexicon.”] The textual world contains more than the surface text, and is not held together in the same ways. Cohesion {39} presupposes coherence (Morgan & Sellner, 1980). Discourse may be cohesive, but not fully coherent (2), or coherent, but not fully, cohesive (3): 

 (2) The slithy toves did gyre and gimble in the wabe. (Lewis Carroll)

 (3) As far as I know / no one yet has done the / in a way obvious now and interesting problem of // doing a / in a sense a structural frequency study of the alternative // syntactical / uh / in a given language / say / like English / the alternative/ / uh / possible structures (Charles E. Osgood) [Osgood (1963: 739) identifies this speech as his own. I translate his notation into mine as best I can; “/” marks a short pause, and “//” a long one. See IV.2.14-34 for a discussion of pauses in text production.

 Some mental disorders (aphasia, schizophrenia, etc.) may impair cohesion but not coherence, or vice-versa (cf. Gosnave, 1977; Rochester, Thurston, & Rupp, 1977; Luria, 1979).

  4.11.3 INTENTIONALITY subsumes the text producer’s attitude that the presented configuration is to be regarded as a cohesive and coherent entity, RELEVANT to (i.e. affecting the chances of) the PLANS and GOALS of the producer. A PLAN is a set of steps intended to lead to a goal; a GOAL is a state of the world whose attainment is envisioned (and usually desired) by the planning agent (III.2.6ff). These notions are central to the domain of “pragmatics” (I.4.4).

  4.11.4 ACCEPTABILITY is the text receiver’s attitude to treat the presented configuration as a cohesive and coherent entity, relevant to the receiver’s plans and goals. Both intentionality and acceptability have THRESHOLDS OF TOLERANCE that apply when the standards of cohesion or coherence are not fully maintained, as in (2) and (3). Plans and goals make a text useable and understandable, even if it would seem disordered or meaningless in isolation. Relevance determines what possible meanings deserve consideration (I.4.6; III.2.33).

 4.11.5 SITUATIONALITY subsumes all the ways in which a text is relevant to (affects the evolution of) a real or recoverable situation. Within the situation, a spoken text constitutes an UTTERANCE ACTION, and a written text an INSCRIPTION ACTION, where ACTION is defined as an event enacted by an agent to change a situation.² [Some philosophers and logicians stipulate that there must be an intention for every action (cf. papers in Rescher [Ed.], 1967). However, a certain margin of discrepancy between intention and action (cf. I.2.8.2; III.3.2.5) doesn’t disqualify many everyday discourse actions. The logically clear notion of “intention” is operationally very complex] One’s reaction to a situation is normally MEDIATED by one’s model of what’s going on (I.1.7). MONITORING occurs when the text serves mainly to give an account of the situation; MANAGING occurs when the text serves mainly to guide the evolution of the situation toward one’s goals.

 4.11.6 INTERTEXTUALITY subsumes the ways in which the text presupposes knowledge of other texts. When the text is remote from others in time or topic, this factor is also substantially MEDIATED. For example, {40} dialogue is usually less mediated than a scholarly citation. Whenever the PRESENTER of a text is not identical with the PRODUCER, e.g., when a passage is quoted or a stage play is performed, or the AUDIENCE is not identical with the ADDRESSEE, e.g., when a personal message is intended for later publication, the current text can be mediated via the situation of its original occurrence.

 4.11.7 INFORMATIVITY is the extent to which text events are uncertain, new, or surprising. In cybernetic terminology, informationality is the degree to which an event disturbs the STABILITY of a textual system and requires REGULATION (I.4.9). Informativity can be operationally subdivided into FAMILIARITY: the extent to which some item or operation has been encountered by the processor; and INFORMATIONALITY: the extent to which any part of a text is unpredictable in view of the rest (III.3.2.8f).1 An INFORMING ACTION presents predominantly new knowledge, whereas an INVOKING ACTION presents predominantly familiar or given knowledge (III.2. 10).

 4.12 Cohesion has been the principal concern of text linguistics (for survey and sources, see Beaugrande & Dressler, 1981; Beaugrande, 1983). Cohesion is the most obviously “linguistic” (language-centered) aspect of textuality. However, the text is not language alone: it is defined as a communicative manifestation in a social and psychological context. Researchers who claim to study language by itself (I. 1. 3; II. 1. 6f, 14; II. 3.7) are in fact supplying the context themselves. Any statements about language in the abstract are ultimately derived from experiences with texts and discourses. Consequently, an interdisciplinary forum is now emerging for an encompassing exploration of texts in human interaction, as illustrated by new journals, e.g.. Discourse Processes (Ablex) and Text (Mouton). The other aspects of textuality besides cohesion are being increasingly emphasized by research from cognitive psychology, ethnomethodology, sociology, anthropology, neuroscience, artificial intelligence, philosophy, semiotics, and cybernetics. The unification of these disciplines gives rise to cognitive science, represented by the journal of that name (also published by Ablex) (and cf. Norman [Ed.], 1981).

 4.13 A fundamental step is to distinguish between an OPERATIONAL approach and a LOGICAL one (cf. II.3.5; III.1.2.8; III.1.8f, 30; III.2.14, 31,40; III.3.13; IV.I.5; IV2.53ff, 66ff; V3.25; VI.3.19). The former encourages us to cut the pie in different ways from the latter. In the traditional philosophical or logical perspective, the investigator’s own reasoning (classifying, categorizing, asserting) was the primary mode of discovery. For example, language was cut up into the levels of sounds, forms, and meanings, and each level had its own taxonomy (I.4.3). Theories were advanced and {42} judged on their logical appeal (completeness, consistency, elegance, etc.). In the empirical or operational perspective, on the other hand, discovery hinges upon seeing how something works and building a model of that. Theories are advanced and judged on their ability to account for real events. For example, the levels of language must now be shown to have a real and distinctive identity in communicative and cognitive processing (III.2.32-40); and a taxonomy of sounds, words, phrases, sentences, or concepts should follow the differentiations humans make when using language. (For instance, whether “modifier” is one operational category or is subdivided into “adjective,”  “adverb,” “participial,” “prepositional phrase,” etc., depends on whether and when people make the distinction.) Operational models are harder to design, but create better conditions for rational adjudication and cumulative progress.

   4.14 What a language entity is—a text, paragraph, statement, sentence, or whatever — is thus defined not by abstract “features” or “markers,” but by what people intend it to be. For the native speaker, the grammar of a language is a set of procedures for using the language, not a theoretical classification of words and patterns (V 2.17, 27). The synthesis of grammatical formats is always presupposed before any linguistic analysis can be undertaken (cf. 1.4.3); and these two perspectives might have quite different priorities. An element is classified not by what it is in the linguist’s scheme, but by what it does, i.e. by its function in the language user’s processing. A single element may have several functions, and a single function may be assigned to several elements. For example, the “linearization” of a text is not the concern of an autonomous level of syntax; it reflects both phrase formats and the requirements of other levels: goal-planning ideation, conceptual development, and sound/letter linearization (cf. IV.2; VI.2). The fact that words are in linear order must now be explained in terms of human intentions and actions that put them in order.

   4.15 The strict logical distinctions of older theories (cf. 11.3.4) often prove to be operational gradations. Depending on context, one aspect or classification may dominate others without by any means excluding them (III.1.17, 24, 29; III.2.5, 10). Operations adapt to their contexts of use, and may redefine language elements by repeating, intensifying, or innovating. Such cases usually encounter a TRADE-OFF. if the gradation reaches an extreme, its categories or results revert to the opposite extreme. For example, making a text very rewarding to read can mean making it very hard to write (III.3.16, 42; IV2.67; VI.I.25; VI.2.37). A balance must be maintained between undue extremes on such gradations as: fast vs. slow, precise vs. fuzzy, certain vs. uncertain, general vs. specific, explicit vs. implicit, elaborate vs. plain, changing vs. repetitive, new vs. given or expected, innovative vs. commonplace, large chunks vs. small ones, and so on (cf. III.1.15; III.2.25; III.3.4.1-10; IV2.38, 43; IV.3.23, 28; VI.I.34; VI.2.2). What is or can be done at any moment depends on what else is going on and how much resources are available (111.3.2ff). Psychologists distinguish a {42} curvilinear function (a rising, then falling slope) from a linear one (a continually rising slope) and an asymptotic one (a rising slope that levels out and stays constant) (cf. Fiske & Maddi, [Eds.], 1961; Estes, 1964). Thus, operations are always subject to the conditions obtaining in the contexts where they occur. For instance, whether you want to repeat or restate something depends on how unfamiliar or complex your materials will seem to the audience: you must navigate between the extremes of confusion vs. boredom.

 4.16 Unlike a logical category, an operation may be degraded or may fail on a particular occasion. A PROBLEM can be defined as a sequence of system states whose passage from the initial state to the goal state might not succeed (cf. Newell & Simon, 1972; Winston, 1977). The problem space is the set of all relevant states that may be traversed. As the probability of failure rises from low to high, the problem goes from trivial to serious (111.1.17). Since the production and reception of texts demands the assembling of systemic options into configurations (1.4.5), these processes qualify as PROBLEM-SOLVING (Beaugrande, 1980a). Depending on context, the seriousness of the problem posed by any operation may call for varying quantities of resources. Detailed research could show how hard text processing is under given conditions, and what sorts of breakdowns or errors are likely to result (cf. III.1.17-20; III.3.1-4). Such inquiries would be an important complement for the less formal view of text production as problem-solving (cf. Rohman, 1965; Young, 1968; Berthoff, 1971, 1972; Larson, 1972; Odell, 1973; Lundsteen, 1976; Coe & Gutierrez, 1981; Flower, 1981).

 4.17 Though the transition from logical to operational language research is far from completed, some implications will be pursued in the following chapters. A theory of text production should be specified within theories of cognitive processing and human interaction. Issues such as resources, memory storage and search, scheduling, motor actions, feedback, attention, motivation, and many more, deserve consideration. The present volume is intended to consolidate some findings and proposals into a research program. To obviate the many disclaimers I should make later on, I want to stress now that the theory, research, and practice presented here (especially my own contributions) should be considered provisional. Many empirical results were drawn from pilot studies. Much recent work is necessarily exploratory, not assertive; approximative, not formal; qualitative, not quantitative; and probabilistic, not law-like. I have sometimes gone beyond the intentions of the original researchers in drawing implications or suggestions from their work. However, I hope that future data-gathering will be done within a continuing discussion of the design of theories and models in evolution. Because of its key role in the whole enterprise, teaching qualifies as research. Theories must never lose sight of the pressing need to promote and guide literacy as a social and educational concern. Though I cannot claim to offer up the “right answers” about text production and composition, I would be more than content if I can help out toward asking some of the right questions.

*    *   *    *   *    *   *    *   *    *   *    *   *    *   *    *   *    *

Click here to go to Chapter II

Click here to go to Main Page