Discourse Analysis and the Art of Coherence Author(s): George Goodin and Kyle Perkins Source: College English,

Vol. 44, No. 1 (Jan., 1982), pp. 57-63 Published by: National Council of Teachers of English Stable URL: http://www.jstor.org/stable/377199 Accessed: 20/02/2009 16:52
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ncte. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

National Council of Teachers of English is collaborating with JSTOR to digitize, preserve and extend access to College English.

http://www.jstor.org

George Goodin and Kyle Perkins

Discourse

Analysis
Coherence

and

the

Art

of

If we may briefly define composition, in many arts, as the arrangement of information, then rhetorical composition is preeminently the arrangement and sequencing of sentences, and the art of coherence, or bonding those sentences together, forms much of the art of rhetoric. Yet most of us have experienced some difficulty in teaching coherence, and the help available in textbooks often amounts to little more than a list of transitionwords. To be sure, we have our conventional wisdom on the subject. For the invention stage of prewritingwe may urge students to make coherence easier for themselves by planningto use the naturalorders of descriptionor narration.For writingitself, the actual stringing together of sentences, we can smooth their paths by suggesting that they concentrate on what they are saying and not permit themselves to worry over spelling, punctuation,and other matters best taken care of later. We might invoke Hemingwayand advise them that when they stop and wonder what to say next they should look at their previous sentences, instead of out the window or at the ceiling. For the postwritingstages of proofreadingand revision, we can adapt Aristotle's method for testing literaryworks, telling them to be suspicious of sentences which can be removed or relocated without any effect on the composition as a whole. It was in the hope of enlarging,however modestly, on such conventionalwisdom that we began to investigate coherence in the light of modern discourse analysis, which is the study of grammatical principlesoperatingbeyond the senitence.Discourse analysis has drawnfrom a numberof disciplines, as we can see in some of the more familiar books in the field: Teun A. van Dijk's Text and Context (1977) draws heavily on semantics and logic, James L. Kinneavy's A Theory of Discourse (1971) on rhetoric, and M. A. K. Halliday and Ruqaiya Hasan's Cohesion in English (1976) on linguistics. Discourse analysis has been

George Goodin is an associate professor of English at Southern Illinois University at Carbondale. A former director of the composition program, he teaches courses in composition, Victorian fiction, and linguistics. Among his publications is an article on literary criticism which appeared in College English in December 1967. Kyle Perkins is an associate professor of linguistics at Southern Illinois University at Carbondale. Among other courses he teaches EFL composition and grammar. He is a former chairperson of the linguistics department and has published in the field of language testing. A version of this essay was presented at the 1981 CCCC meeting in Dallas.

College English, Volume 44, Number 1, January1982 57

58

College English

used to study not only written texts but spoken monologues and conversations as well. It has studied how rhetoricalform or macrostructure governs the writer's syntactic options, and how these in turn help the reader to grasp rhetorical form and to understandcompositions as wholes. It has studied how the reader's habits of informationprocessing constrainthe writer's selection and orderingof details. It has studied how fictional narrativesmove from one given episode to another with considerableregularity. Althoughdiscourse analysis must by its nature deal with coherence, this fact by no means makes it all equally useful to teachers of composition. A typical bit of discourse analysis cites a coherent passage and explains in some detail the bonds among its sentences, just as a typical bit of grammaticalanalysis cites a well-formed sentence and explains its internal bonds. Unfortunately,discourse analysis, unlike the grammarof the sentence, does not have a large body of well-specified and obligatory rules which utterances not well formed may be seen as violating. Thus composition teachers may well be disappointed. They bemay feel quite comfortablealready on the subject of coherence, particularly cause the poetics of the last forty years has used it as the primarystandardfor judgingliteraryworks. Whatthey more likely want is a better handleon incoherence.

For this purpose the most helpful work in discourse analysis that we have found is Herbert H. Clark and Susan E. Haviland's "Comprehensionand the Given-New Contract" (in Roy O. Freedle, ed., Discourse Productionand Comprehension, [Norwood, N.J.: Ablex Publishing, 1977], pp. 1-40). We can hardly do justice here to their very rich, rewarding, and-we are happy to sayreadable article, but briefly their argument goes as follows. Each declarative sentence contains both "given" informationand "new" information.The given informationis that which the speaker believes the listener alreadyknows, either from the surroundingsor from previous sentences, and the new informationis that which the speaker believes the listener does not know and needs to be told. An implicit agreementbetween speaker and listener, which Clarkand Haviland call the "given-new contract," governs how each of these is manifested in a sentence. Essentially the given informationis manifested by unstressed intonation and/or sentence position, and the new informationby stressed intonation and still not get and/or sentence position. Perhaps we can assist understanding bogged down in a difficultand disputed matterby suggestingthat the unstressed position for given informationis in the logical (not necessarily the grammatical) subject, and the stressed position in the logical predicate. Those familiarwith linguistics will thus recognize that Clarkand Haviland'sdistinctionis analogous to a whole battery of previous distinctions between topic and comment, presupand new information. position and focus, theme and rheme, and old information As to sentence sequences, the rule is that once informationhas appeared as new information,it becomes for subsequent sentences available to be used as given information. For each subsequent sentence, then, the speaker observes what Clarkand Havilandcall the "maxim of antecedence," which they state as follows: "Try to construct your utterancesuch that the listener has one and only [one] direct antecedent for any given informationand that is the intended an-

Discourse Analysis and the Art of Coherence

59

tecedent" (p. 4). When this maxim is violated, two results may occur. One is unacceptability,which they illustrate with the second of the following pair of sentences: "Two men were watching the dog. The one watching it laughed out loud" (p. 16). The other is awkwardness,which they illustratewith the second of the following pair: "Agnes saw somebody. It was Agnes who saw Maxine" (p. 14). The difference between unacceptabilityand awkwardness is that listeners can often comprehend awkward sentences by making them conform to the maxim of antecedence, using processes of inferentialbridging, addition, or restructuring.The last example, if restructuredto remove its awkwardness, can become "Agnes saw somebody. It was Maxine whom Agnes saw." Now, how can we use this very interestingdiscussion of faulty sentence sequences? We can begin by noting that all sequences which Clark and Haviland find unacceptable or awkward may also be called incoherent. As a result, we have a good diagnosis of incoherence which violates the maxim of antecedence. For example, the following sequence from student writing requires inferential bridgingby the reader: "In the past ten years many more young people have realized that their future in this advanced technological society lies in attending college. Among this growingmass of test takers are three times as many students from generally low-scoring groups." Some sequences, however, observe the maxim of antecedence and are incoherent nonetheless, as in the following pair, used to open a student paper: "Many people have never taken the opportunityto play a card game. Card games seem easy after learning the rules." To process the second of these, the reader must also provide inferential bridging, to the effect that many of the people who have never taken the opportunityto play card games may have been deterredby thinkingthem difficult. this problem,we think Clarkand Haviland'swork very helpNotwithstanding ful for analyzingincoherence. It suggests that we look for rules of coherent discourse which are obligatory and therefore permit us not only to diagnose incoherence, but also to specify more exactly than we have in the past the sentence which violates coherence. By showing that discourse develops by means of a serial or recursive process-converting new informationinto given informationthat forms the base for predicatingadditional new information-it also suggests that the art of coherence may lie in such acts of conversion. If discourse works in the way we have described, there are two obvious ways in which it can go wrong. A sentence may have little or no informationthat is really new, but may simply repeat or rephrase what has preceded. Discourse greatly markedby this defect is called by our students "bullshit," and many of them can fill blue book after blue book with it. On the other hand, a sentence may be deficient in given information. If so it is incoherent with its context. Discourse greatly markedby this defect may leave us wondering what is being talked about, or it may sound like a collection of topic sentences. In describing incoherence in this way, we have greatly broadenedthe maxim of antecedence, and admittedlyour reformulationlacks the precision and elegance of Clark and Haviland'srule. Nevertheless, we think it does point in the right direction. We wish we could provide from discourse analysis a more helpful analysis of incoherence. Perhaps people working in the field will some day create one, but

60

College English

we must be realistic and not get our hopes too high. After all, discourse analysis has been going on for centuries in the work of logicians and rhetoricians, and the entry into it of linguistics and other disciplines cannot be expected to banish muddle. Just as we cannot get students to write consistently grammatical sentences by teaching them any system of grammar, we do not expect to get them all to write consistently coherent prose by teaching them any system of discourse grammar or rhetoric. We want to suggest not that composition teachers study a great deal of discourse analysis, and certainly not that they wait for it to solve their problems, but rather that, with a minimum of terminological baggage, they do discourse analysis (or continue to do it) with their students. In our classes we have treated the grammar of the sentence unsystematically by presenting students with their own ungrammatical sentences, relying on whatever level of competence they already have to spot the parts that sound odd and to correct them. Similarly, we have presented them with copies of their paragraphs containing sentences which violate discourse coherence, relying on the competence they have to sense that something is wrong and to suggest possible revisions. Some of this material can be found in the writing samples which we want to consider now, much in the way that we did with our students. Here is the first sample: 1. There are several differencesbetween Perrault'sversion of the story Cinderella, and the newer version of the story by Walt Disney. This paragraph,however, will discuss only one of them. This difference is the fact that in Perrault'sversion, Cinderellais resigned to the fact that she cannot go to the Prince's ball because of her position in society. She felt, as her sisters did, that it would not be acceptable to the people giving or attendingthe ball. However in Walt Disney's version of Cinderella, Cinderellafelt that she should be able to go to the ball if she wanted to, and she was angry at her sisters and stepmotherfor not allowingher to go. I feel this differencewas probablybroughtaboutby people's changing attitudes toward being caught up in social classes. Back in Perrault's day, people felt confined by their class from the day they were born. (I realize Cinderellawas not born into the class of a servant, but her stepmotherreduced her to that level in the way she had her familytreat Cinderella.)In WaltDisney's more recent writing of Cinderellathough, there is shown the difference in that people today do not feel confined to their social class. People from lower social classes feel they may go almost anywhere without feeling unaccepted. This is one difference I found between Perrault'sCinderella and Walt Disney's Cinderella .

Our first sample is a quite coherent paragraph. In fact it inclines toward the opposite error of repeating old information too much: many phrases and perhaps a sentence or two can be omitted without much damage. Nevertheless, the paragraph is quite good. After the first sentence we can see many instances of the discourse attaining movement and coherence by treating what is new information in one sentence as given information in a subsequent sentence. The "several differences" and so forth of sentence one is new information, but in sentence two, where it is referred to by "them," it is given information. The new information of sentence two, that "this paragraph . . . will discuss only one," contains in the word "one" new information which will be treated as given information in the expression "this difference" of the following sentence. Further analysis would be tedious. As we said before, showing that coherent writing is coherent

Discourse Analysis and the Art of Coherence

61

is not the main problem, and this writer is conscious enough of coherence and knowledgeable enough of convention to put into parentheses an entire sentence which would have violated coherence if not marked as an aside. Such consciousness does not prevail in our second sample: 2. I think most people watch soap operas to take themselves away from reality. When you really get interestedin a soap opera, you get involved in the world of characters.You relax for awhile and enjoy the story. *Soap operas are just exaggeratedstories. It is easy to forget your problemsfor one-halfto two hours of the day watchingthe characterswith their innumerable problems.These characters are portrayedlike real people. Some of their problems could possibly be real. In the second sample the asterisked fourth sentence is an aside which is unmarked and therefore violates coherence. If speaking it the student might very well mark it as an aside by lowering pitch and increasing speed, but he or she has no equivalent graphic symbols. (If we may interject an aside of our own, many writing problems are similar-not caused by ignorance of language itself but of its limitations in certain circumstances.) In any case, the fourth sentence does not violate the maxim of antecedence, but it introduces incoherence because it requires the reader to make an inferential bridge to the effect that the exaggeration in soap operas is a source or the source of the pleasure some people receive from them. We continue the discussion of unmarked asides with our third and fourth samples: 3. The card game Thirty-One is played with a card deck containingfifty-two cards. *For an interestingand challenginggame the numberof players can range from two to six. The cards have face value with face cards equalling ten and aces equallingeither one or eleven. Each player is holding three cards at the end of each turn. The object of the game is to have the sum of the values of the three cards equal to thirty-oneor lower than thirty-onewithout ever going over. 4. War is a card game set up for two or more players. All fifty-two cards from a standarddeck are dealt evenly to each player. The player with all the cards at the end of the game is the winner. *Thisgame is played mostly by young people. In the third sample the asterisked second sentence is also an unmarked aside. If it is removed the paragraph becomes coherent enough. We are not suggesting that it should be removed, or marked as an aside. Instead, its essential information should be embedded in the previous sentence, which might then read, "The card game Thirty-one is played, ideally by two to six people, with a deck containing fifty-two cards." Similarly in the fourth sample, the first sentence can incorporate the information contained in the last sentence if revised to read, "War is a card game set up for two or more players-usually young." The problem is more serious in the fifth sample: 5. One unique difference of the versions of Cinderella is the length of the story. *Walt Disney's object of the story was to simplify the understanding for children. Perrault'sform of Cinderellaexplaineddetailed informationthat was constructed in a more advanced and detailed manner. The styles of the writings contrastedgreatly in this type of formality.Even though Walt Disney wrote in a brief style, the story still revealed the same plot as Perraultdescribed in his
version.

62

College English

Here the offending sentence is by no means an aside. The first sentence should make the reader expect soon information as to what the "versions" are and which is longer or shorter. The next two sentences identify the versions, fairly well satisfying the maxim of antecedence, but they fail to deliver the information about length that is necessary to understand any assertion about the purpose or object of each version. This deficiency in given information would never have arisen if the first sentence had read, "One difference between Walt Disney's Cinderella and Perrault's is that Walt Disney's is shorter." Thus revision to eliminate incoherence can address itself either to the offending sentence or to the sentence setting up the unmet expectation. In this respect coherence problems, like many grammatical problems within sentences, are matters of concord. Concord also is the issue in the sixth sample: 6. When Alice met the pigeon; the pigeon thought she was a serpent. One reason for this identity problemis because there are many ways to perceive an object. *Alice tried to explain to the pigeon that she was a girl, but the pigeon had seen many girls, and none had such a long neck. When Alice told him that she ate eggs, the pigeon was convinced that she was a serpent. Anythingthat eats eggs was a serpent as far as the pigeon was concerned. The same object viewed throughdifferenteyes, can be perceived in manydifferentways, therefore,what may be a sweet little girl to some, maybe a serpentto others. In sample six the asterisked third sentence does, after a fashion, begin to develop the idea in the second, but not clearly. Notice, however, that the paragraph becomes quite coherent if the second sentence is removed. The general observation it makes is also in the last sentence, which is made intelligible and coherent by what precedes it. By then, we can recognize that "the same object" is in this case Alice, and the "different eyes" those of Alice and the pigeon. Removing the second sentence makes the paragraph a simple narrative with its point at the end. Here is the final sample we wish to discuss: 7. The finally, Japaneserespect the Americanpeople in some points but not at all. I know some crazy Americanpeople as same as Japanese. If the Americanpeople want to come to Japan, act carefully. It is a cause of misunderstanding. This passage obviously was written by a student of English as a second language. Its individual sentences, except for the last, are so incoherent internally that bonds between them have become difficult to make. They must be reprocessed by the reader. The first sentence must have "at" subtracted. The second must be restructured. The third must have "Americans" substituted for "the American people" and must be restructured to remove the shift from third person to second. Only then will the discourse problems of the paragraph come into view. These problems, unlike those of the preceding samples, are with what Halliday and Hasan call "cohesion." Before the student can deal with coherence proper, he or she must recognize the unnecessary shift of subject from the first to the second sentence, and the semantic and cohesive properties of the definite article in English.

Discourse Analysis and the Art of Coherence

63

These samples and many others have suggested to us some tentative comments or generalizationswhich others may wish to consider as they work with discourse coherence. First, structurewithin the sentence obviously has a great impacton coherence between sentences. In samples two throughfour, the offendingsentence ought to have been embedded in another sentence instead of being allowed to stand independently. The coherence problems, therefore, may have originatedfrom a lack of syntactical maturity in the writers, which is much in evidence throughout these samples. The sentences which introduce incoherence may in these cases be thought of as semantic analogues of those sentence fragmentsreferredto by composition textbooks as "afterthoughtmodifiers." A second observation also suggests an analogy with afterthoughtmodifiers. By far, the most common location we have found for the offending sentence is second in a paragraph,particularlythe second sentence of the first paragraph. The first sentence is often a thesis or topic sentence almost inevitablydominated by new information,and it may quickly become so meaty as to tax the writer's syntactical competence. If so, he or she may simply add as an independentsentence materialthat should have been embedded, and thereby violate coherence. The next most common location for improperlybonded sentences is the last sentence in a paragraph, and here againmaterialwhich needed embeddingearlier is often simply stuck in. Lest these remarkson syntactical immaturityseem to imply that exhaustive work on sentence-combiningshould precede any serious instructionon intersentential coherence, we hasten to a third observation. Macrostructure,the paragraph or the composition as a whole, also has a great impact on coherence. Many of our samples show that the writer has formed no definite idea of what purpose a paragraphis to serve or what means will be used to develop it. The sixth sample, the paragraphon Alice, as we have seen, suggests in the second sentence that development will be by analysis; then it shifts to narrative.The writer seems to think that a paragraph ought to have an early topic sentence but seems to be unaware that most topic sentences, particularlyearly ones, create the expectation of development by analysis-the most difficult means of development that could possibly be used. Our last observation, which we think is amply warrantedby the differences which our samples show between native and non-native speakers, is that native speakers-fortunately for those who teach them-already know a great deal about coherence. For the last twenty years or so at least, composition teachers have been outragedby the methodologicalassumptionof linguistics that native speakers already have grammaticalcompetence, and they can certainlypoint to a great deal of incompetence. But the fact is that native students do have a great deal of grammatical competence, and we would do well to build on it, pointingit out to them as somethingto be proudof, ratherthan assumingthat we are working with a tabula rasa. And just as we can use their grammaticalcompetence as our startingpoint for trying to increase it, so their discourse competence should be our starting point for extending and deepening their knowledge of the art of coherence.