Skip to content

Composition Forum 49, Summer 2022

The Discourse-Based Interview: A Procedure for Exploring the Tacit Knowledge of Writers in Nonacademic Settings

Lee Odell, Dixie Goswami, Anne Herrington

Editors’ note: We are including this reprint of “The Discourse-Based Interview” in the special issue because many researchers have observed that the original text can be difficult to find, given that it is out of print. Lee Odell, Dixie Goswami, and Anne Herrington gave their permission to reprint the chapter here, as did the surviving collection editors, Lynne Tamor and Sean A. Walmsley. We thank the authors and editors for this generosity.

This web-based version duplicates the text as it appeared in the original edited collection, without pagination and with minor formatting adaptations necessary for web presentation in Composition Forum. We prepared this manuscript by scanning a clean copy of the original chapter, using optical character recognition to convert it to text, correcting mistakes introduced by the process, then reading the chapter out loud to ensure accuracy. We have also prepared a clean PDF that duplicates the original pagination, for researchers who prefer to cite page numbers.

Originally published in Research on Writing: Principles and Methods, edited by Peter Mosenthal, Lynne Tamor, and Sean A. Walmsley, Longman, 1983, pp. 221-236.

According to Michael Polanyi (Polanyi & Prosch, 1975), much of our knowledge is personal and tacit. We acquire this knowledge not so much by memorizing rules or reading textbook explanations as by repeatedly engaging in a given activity or, in Polanyi’s terms, “dwelling in” a particular action. For example, we develop our skill as writers not by studying rules, but by continually writing. Further, it is likely that we do not consciously formulate much of this knowledge as a set of premises or maxims, but instead internalize it as inexplicit functional knowledge that we shall use and expand upon each time we write. Polanyi argues that this knowledge is characteristic of all activities, whether physical—such as riding a bicycle—or mental—such as solving a difficult chess problem. Polanyi also claims that tacit knowledge exists at a number of levels. Most significantly for our purposes, Polanyi cites the example of oral communication. At the lowest level, we have learned a phonological system (although most of us could not readily explain how it works), and at the highest level, we have learned to become sensitive to rhetorical or interpersonal context; we have inferred from experience how to vary our style of expression according to the audience we are addressing. This knowledge includes our understanding of both the contexts in which we speak and the strategies that are appropriate for a given context.

A number of sociolinguistic studies of oral language have confirmed the influence of social context on speech. In a review of research on language acquisition, Cazden (1970) stresses the situational relativity of children’s speech and cites research evidence that shows the influence of such situation-specific factors as topic, task, and speaker-listener relationship. Labov (1970) cites the influence of these same factors on stylistic shifts, and he asserts that “there are no single-style speakers... every speaker will show some variation in phonological and syntactic rules according to the immediate context in which he is speaking” (p. 16). A similar conclusion appears to hold true for writing. Discourse theorists and teachers of composition (Booth, 1975; Gibson, 1969; Kinneavy, 1980) argue that writing does not exist in a vacuum, that a writer’s purpose and knowledge of audience and subject shape the stylistic and substantive choices the writer makes. This point of view receives some support from several recent studies that show that certain groups of writers vary syntax according to the rhetorical context, the audience, and purpose for which they are writing (Crowhurst & Piché, 1979; Rubin & Piché, 1979).

By studying the ways that rhetorical context influences writing, researchers have begun to confirm assumptions that are widely held but that have not been subjected to careful testing. Moreover, results of these studies help justify pedagogical and evaluative practices recommended by Lloyd-Jones (1977), Moffett (1968), and others. However, studies of written language are limited in two respects. For one thing, composition researchers have carried out their work in classroom or experimental settings. With few exceptions (e.g., Scribner & Cole, 1978), researchers have not studied the writing people do as a part of their daily lives in nonacademic settings. Moreover, few researchers have attempted to understand the tacit personal knowledge that writers bring to bear on their writing tasks. Thus, existing research tells us, for example, that the syntax of student writing addressed to a teacher may differ from the syntax of student writing addressed to a close friend. But this research is not likely to help us understand the tacit knowledge the writers brought to bear on these tasks. We cannot determine what assumptions writers made or what background knowledge they had concerning the audience, the topic, and the strategies that might be appropriate for achieving their assigned purpose with a given audience.

These limitations seem important. We know (Goswami, 1978; Van Dyck, 1980) that some workers in nonacademic settings frequently have to write for diverse audiences and purposes. We have reason to think (Knoblauch, 1980; Odell & Goswami, 1982) that some of these workers possess detailed, useful information concerning the occupational and rhetorical context for their writing. We believe that much of this information may be tacit knowledge. That is, having derived it through repeated experience, writers can use it without having to formulate it consciously each time they write. We also believe that this knowledge may be of interest to both theorists and teachers. Consequently, we want to raise a methodological question that will occupy the rest of this chapter. How can researchers get at the tacit knowledge of people who write in nonacademic settings? What methodology will enable writers to make explicit the knowledge or strategies that previously may have been only implicit?

Our answer to these questions is illustrated in the following letter from a business executive to a sales representative. This letter, part of a larger sample collected from this executive, has been modified so that at five points it indicates both the original text and an alternative the writer might have chosen. In three of these instances (1, 4, and 5), the writer was asked to consider using an alternative form for each of the following: addressing his reader (1), asking the reader to perform an action (4), and referring to himself (5). At two other points, the writer was asked to consider deleting an introductory, context-setting statement (bracketed passage at 2) and a passage that elaborates on a general term (bracketed passage at 4). In all five instances, the alternatives were, in fact, reflected in some other piece of this executive’s work-related writing. To elicit information concerning the writer’s tacit knowledge about the rhetorical context for the letter, an interviewer asked, in effect, two basic questions: “Here you do X. In other pieces of writing, you do Y or Z. In this passage, would you be willing to do Y or Z rather than X? What basis do you have for preferring one alternative to the other?”

When asked about the first alternative, the form used in addressing his reader, the writer was not willing to use “Dear Mr. Bunch” rather than “Dear Ron”:

[“Dear Mr. Bunch”] is a possible alternative, but I was trying to establish with “Dear Ron” that I’ve talked to him a number of times and I feel that we’re on some sort of a personal basis and that’s what I was trying to establish. It’s a business letter but I didn’t want to make it so stiff.

The writer also declined to omit the context-setting phrase, “Pursuant to our conversations over the past few months...”

I do want to get the point in about the fact that we’ve talked about this a couple of times in the past. He was, quite frankly, chasing the daylights out of me to get this account. We’ve been talking on and off and it’s been generally at his initiative and, now that we’ve made a positive decision, I want to recall that to him, if you will, in such a subtle way as to further make him do the job.

In these two excerpts from his interview, the writer gives us information about his actual and his desired relationship with Ron Bunch. He reveals that he has had frequent personal contacts with Ron Bunch, contacts that Bunch had initiated. His comments also suggest he is trying to maintain a rather delicate relationship with his reader. The writer wants to avoid the formality that characterizes some employer-employee relationships, yet he retains the rather authoritative role of someone who is responsible for seeing that another carries out a job as effectively as possible. Furthermore, the writer’s comments imply at least two rhetorical strategies. The first is commonplace enough, that is, the writer uses the reader’s first name to establish or confirm a personal relationship and to create a tone that is not too stiff. The second strategy seems somewhat less obvious. The writer subtly reminds the reader of his previous expressions of interest to enhance the reader’s motivation.

27 September 1979

Mr. Ronald R. Bunch
Marketing Corporation
100 Southward Island
Clearwater, Florida 33500


(1) Dear Ron:
Dear Mr. Bunch:

(2) [Pursuant to our conversations over the past few months and in line with our need and desire for professional sales coverage in Florida], I am happy to report that you have been chosen to represent the PDS portion of the Acme Amalgamated product line.

(3) As a result, I have enclosed two copies of our sales representative agreement covering PDS products. [This agreement has an 11/1/79 effective date and you will receive 5% commission on the listed products for all invoices dated 11/1/79 and beyond. This, of course, includes all new orders received on or after this date plus all orders presented in house.]

(4) [Please sign...
You must sign...
It is imperative that you sign...]
both copies of this contract and return one to us for our records.

Ron, it is indeed a pleasure to have you as part of our sales team, and I am excited about the prospects for the future.

(5) [I am looking forward to...
We are looking forward to...
Amalgamated Products is looking forward to...]
a long and mutually beneficial relationship.

If there should be any questions in this matter, please call me.

Sincerely yours,

J. F. Moon
Product Manager


In discussing the alternatives for paragraphs three and four, the writer expresses his sense of what he may and may not assume to be true of his reader. He is unwilling to delete the details about the sales agreement (3), in part because he cannot be sure about his reader’s prior experience:

I want it understood up front what we are going to pay commissions for and when we are going to start paying the commissions. The important part of any relationship is the beginning. And I don’t want anything inferred or assumed. I want the facts clearly stated.... [This arrangement] may be different from what he has been accustomed to in the past.

Yet, in discussing ways to phrase a request (4), the writer assumes that he and his reader share a certain amount of common knowledge. He refuses to accept the alternative, “It is imperative...,” because he believes that the reader will recognize the importance of the request the writer is making:

Obviously, it is imperative that he sign both copies but I don’t think I would choose that particular type of phrasing for it. We’re dealing with a professional sales representative and we’re supposed to be professionals; it’s implied that both copies have to be signed before it’s valid. “Imperative” would seem like too strong a word in this particular context....

A moment later in his interview, the writer reveals a strategy for evaluating an alternative, a strategy that entails asking himself how he would react to a given phrase in a particular context (4):

If I heard that, “it is imperative that you sign...” something that obviously is going to be done, my response would be nervousness, or some other thought.

In commenting on the fifth alternative [the form he will use in referring to himself (5)], the writer seems, at first, to be guided only by a simple stylistic rule:

You should basically stay away from too much of “I, I, I” in a business type of letter. That’s a mistake.

Yet, he immediately goes on to relate his preference for “We are looking forward...” because of his sense of his relationship to his company and his reader. The writer points out that using “I” in this context implies that the reader is:

Dealing one to one with me as opposed to me as a representative of Amalgamated Products as a whole. He understands perfectly well what’s going on here, but [using I] is a little bit presumptuous on my part.

We cannot, of course, argue that the business executive consciously considered the relationship between himself, his company, and the reader while he was actually writing his letter. Although we did not observe the executive’s composing process, we would expect that several of the alternatives mentioned previously were chosen with little or no apparent deliberation. Indeed, because the executive was an experienced writer and this was not a terribly unusual task, we assume that the task of writing this letter was eased somewhat by knowledge that the writer may not have explicitly formulated while writing the letter, knowledge about the subject, the audience, and the strategies that were most likely to prove effective.

Information about this knowledge is interesting, because in part, it may help us test discourse theorists’ claims about ways in which considerations of audience and purpose are important for writers. Furthermore, this information has practical implications. For example, the Ron Bunch letter and interview transcripts are from a study conducted by David Lauerman and his colleagues (personal communication) at Canisius College in Buffalo, N. Y. As part of their effort to design an advanced composition course for students in business, Lauerman and his colleagues have collected an extensive writing sample from executives in several different businesses and have interviewed these executives using the procedures we have described. As a result of this work, Lauerman is able to create writing tasks that actually reflect some of the rhetorical demands students will encounter in their careers. Further, interview materials from the study frequently serve as a basis for class discussion. For example, students are frequently given a piece of writing comparable to the Ron Bunch letter. Students are asked to decide which alternatives seem most appropriate to them and then compare their choices and reasoning with those of the original writer.

To summarize our argument thus far: (1) our interview procedure can be used with writers in diverse settings, writers whose ability may vary widely; (2) interviews with these writers enable them to tell us about the tacit knowledge they bring to writing tasks they encounter every day; (3) information about this knowledge is of interest to both theorists and teachers.

This series of claims raises a number of questions:

How does one justify studying the writing people do routinely? Would we not elicit more information if we designed more challenging experimental tasks that would tax their composing skills more severely than a routine task?

How valid is interview data? Can we have any confidence in the observations that writers make well after a given piece of writing has been completed?

Why should researchers, rather than writers, determine what features of a text are discussed in an interview?

How significant are the features we have selected for writers to comment on?

How does our research methodology relate to the ‘compose-aloud’ procedure used by several other researchers?


Because they are interested in studying writers’ underlying composing processes, such writing researchers as Flower & Hayes (1980), have made extensive use of experimental tasks that are designed to pose unique and unexpected demands for a writer. Their rationale is that to do such tasks, writers would have to draw on their full repertoire of composing strategies and not rely on “stored problem representations’’ that they have developed for routine tasks. Because such experimental tasks are designed to avoid familiar, natural contexts, they are not suitable for our purpose, which is to probe a worker’s store of knowledge of the rhetorical context for writing done on the job. Consider, for example, an experimental task, such as “Write about abortion, pro and con, for Children’s Digest, which is read by ten- to twelve-year-olds.” This task might elicit information about strategies a social services administrator would use to solve unique tasks for which he or she has no context, and perhaps elicit as well information about global strategies the administrator uses. But it would not elicit information about the contextual knowledge that shapes that administrator’s writing on the job or about how global strategies are combined with task and context-specific knowledge to compose a particular piece.


One answer to this question is suggested by Atlas (1979). In reviewing widely reported techniques for studying writing, Atlas concluded that the validity of interviews “depends heavily on the accuracy of the subjects’ self knowledge; for this reason, interview data is probably best treated as weak evidence, suggestive but not conclusive” (p. 36). Our response to this criticism depends on the use to which an interview is being put. If a researcher is using an interview to determine what went on while a writer was engaged in the process of composing, we agree with Atlas. When experience is transferred from short-term memory (STM) to long-term memory (LTM), we assume that it is simplified and restructured; it seems unlikely that LTM can retain the full complexity of mental activity attendant on the moment-by-moment process of composing. Further, we agree with Polanyi and Prosch (1975) that some tacit knowledge is so internalized that it becomes unconscious and inaccessible. However, we are not using interviews to obtain information about mental processes. We are using interviews to identify the kinds of world knowledge and expectations that informants bring to writing tasks and to discover the perceptions informants have about the conceptual demands that functional, interactive writing tasks make on them. Research on verbal reports as data confirms that informants can report reliably on such socially learned information, which has been tacitly transformed into functional plans they apply when writing (Smith & Miller, 1978).

Later in this chapter, we shall argue that our interviewing procedures are particularly well suited for eliciting this sort of information. But the validity of this information is subject to at least two other criticisms. It may be that an interviewer will bias a writer’s response by the kinds of questions he or she asks simply by deciding to ask about one feature of a text rather than some other feature. Further, there is the chance that interviewees will mislead researchers and themselves if only by allowing feelings or preconceptions to influence their statements. There may be no way to satisfy completely the first of these criticisms. The very act of observing any phenomenon may alter that phenomenon. However, we have devised interview procedures that will help an interviewer be as non-directive as possible (Odell & Goswami, 1982). Furthermore, we are inclined to trust interviewees’ statements. For one thing, interviewees rarely respond to our questions with abstract precepts about good writing. Instead, they usually talk specifically about the interpersonal and occupational context in which their writing exists. It may be that any single one of these statements is suspect, that at any given moment an interviewee may mislead himself or herself or the researcher. But we have some evidence that interviewees’ statements seem to vary according to the type of job they hold. We find patterns in the statements of workers in one group that differ from patterns in statements by workers in another group. This blend of consistency within groups and variations between groups leads us to believe that interviewees’ responses are not simply individual writers’ whims or misperceptions, and are not governed solely by our intrusion as interviewers or by some bias in one interview technique.


As we have noted, our interview procedures require the researcher rather than the writer to decide which feature of the text will be discussed. We know, of course, that a finished, edited text gives no clue as to what parts of the text required extensive deliberation and what parts were written quickly with little conscious effort. Consequently, it is quite possible that the interviewer will fail to ask the writer about matters that occupied large parts of the composing process. This possibility raises two further suggestions: perhaps the writer should identify at least some of the matters to be discussed in the interview; perhaps the interview should not be based solely on the finished text. In response to the first speculation, we must point out that we are interested in knowledge that may not be consciously learned or applied. If this knowledge is not at the front of a writer’s consciousness while composing, it seems unreasonable to expect the writer to identify points at which he or she has relied on that knowledge. As to the second suggestion, we agree that there may be times when observation of a writer’s composing process can enable a researcher to ask questions about, for example, points at which a writer made and crossed out several false starts. Or it might be possible to base an interview on the revisions a writer makes in successive drafts of a piece of writing (see Cooper & Odell, 1976). Yet these strategies may tell us only part of what we want to know. Insofar as, for example, revision entails tacit knowledge, we might profit from interviewing writers about their revising. But this knowledge is almost certainly functioning when a writer does not have to stop, deliberate, and revise. Indeed, we believe that the transformation of contextual knowledge into tacit plans is what enables large parts of the composing process to proceed with little conscious effort; tacit knowledge is not limited to those parts of a text that require revision.

This last assertion raises a series of questions. If many parts of a text may have allowed a writer to use his or her tacit knowledge, how will researchers decide about what parts of a text they will ask questions? How can researchers be sure they are not ignoring important parts of a text? What basis does one have for assuming that a given feature of the text is significant?


In trying to decide what parts of a text we would ask writers to comment on, we made several assumptions. The first was that writers, like speakers, are not univocal; that is, they are capable of varying the language, syntax, and content of their writing. Consequently, we gathered samples of writing from each person we interviewed, materials representing the full range of written tasks that the person typically did as part of his or her job. We examined these writing samples, looking for variations, trying to identify the alternatives that were part of each writer’s repertoire. Our next assumption was that many of these alternatives might have been chosen with little or no conscious deliberation, that a particular locution or bit of information may have seemed so routine or so uniquely appropriate that a writer might not recognize it as a choice. Finally, we assumed that if we asked writers to consider alternatives (alternatives that were evident in other materials he or she had written), we might create a cognitive dissonance that would enable a writer to become conscious of the tacit knowledge that justified the use of a particular alternative.

Having identified alternatives to discuss with writers, we still cannot provide a completely satisfactory answer to the second question we raised earlier; we must acknowledge that there may be other features of a text that will provide as much information as, or perhaps more information than, those features we have chosen. As soon as researchers direct their attention in one direction, they blind themselves, at least temporarily, to information that might be available if they were to look in another direction. But we can mitigate this problem if we ask writers about a variety of features in a given set of texts and if we are willing to ask about different kinds of features when we are dealing with different sorts of texts. For example, in a study of writing in a welfare agency, Odell and Goswami (1982) interviewed administrators about their letters and memos, texts that involved many of the same alternatives as were illustrated earlier in the letter of Ron Bunch. However, caseworkers in the welfare agency rarely wrote letters and memos; instead, they were most likely to write reports of their meetings with clients. In interviewing caseworkers about these reports, it was necessary to consider a different set of alternatives. In place of asking about the form used in addressing a reader or the way they signed their name to a letter or memo, the researchers asked about such decisions as whether to refer to a client informally (by just using his or her first name), formally (by using Mr., Mrs., or Ms.), or impersonally (by referring to the client as client); whether to include/exclude information about the caseworker’s actions during a meeting with a client; or whether or not to refer to the client’s actions.

On the face of it, some of the alternatives discussed in our interviews seem rather insignificant. For example, in our interviews based on letters and memos, we have asked writers about the way they signed their name. We assume that writers may spend very little time trying to decide whether they should use their full name or just their first name. Yet questions about such an apparently simple matter as this can provide a great deal of information. When we read a collection of one administrator’s writings, we noticed that she had several different ways of signing her name: M. Smith, Margaret Smith, Meg Smith, and Meg. In one of our interviews we asked her if she would be willing to sign her name on a particular letter as Margaret or Meg Smith rather than M. Smith. Here is an excerpt from her reply:

This [letter] is going to a permanent file. I am looking to the years to come. Someone coming back.... It makes no difference whether I am male or female making this decision. [What matters is that] I am a grade A supervisor. They have to know where he is placed and who evaluated him. But I don’t use Margaret Smith for this reason: I want to be neuter.

In commenting on the way she signed a memo, the administrator remarked that she preferred the signature Margaret Smith (rather than M. Smith) because:

This is not a formal little note.... I’m sharing some information, so I put Margaret. I use that M. when I don’t want the reader to know whether I’m male or female.

In yet another interview, this writer noted that it was unusual for a woman to hold a high administrative post in the agency where she worked. She remarked that she sometimes felt her writing carried more weight when a reader did not know whether the writer was a man or a woman. Thus, our inquiry into an apparently simple matter elicited a great deal of information about the writer’s understanding of her status in the agency and about one of her strategies for accomplishing her work.

We want to make a similar claim for the other alternatives used in interviews with workers in the welfare agency. All of these alternatives elicited information about writers’ knowledge of the rhetorical and occupational context for their writing. Further, writers’ comments about these alternatives reflected writers’ knowledge of ways to vary style and substance to achieve particular effects. For example, the Ron Bunch interview illustrates that the questions about form of address (1), form of command/request (4), and form of reference to self (5) reveal the writer’s knowledge of how to establish the desired writer/reader relationship, the professional context, and the conventions within that context. In short, although an isolated feature (e.g., form of address or elaboration) may seem insignificant, it is a sensitive indicator of writers’ complex understanding of the rhetorical context and ways for them to achieve their purpose within that context.


Flower and Hayes (1980) have shown the usefulness of asking writers to compose aloud, to verbalize the thoughts and feelings that accompany their efforts to complete a piece of writing. Although composing aloud was not used in the study of writers in a welfare agency (cited earlier), we realize that this procedure has enabled some writers to comment on the rhetorical context for their writing. Consequently, we asked four welfare workers (two administrators and two caseworkers, all of whom participated in the study by Odell & Goswami, 1982) to do some composing aloud. One administrator did two composing-aloud tasks. Each of the other participants did four such tasks. As we analyzed their work on these tasks, we gained some understanding of both the uses and limitations of the composing-aloud technique. One argument for the composing-aloud methodology is that it can be an excellent way to get at the generating, planning, and organizing activities that make up a large part of a writer’s composing process. Moreover this procedure sometimes can give a good record of what Britton has called, “shaping at the point of utterance” (Britton, Burgess, Martin, McLeod, & Rosen, 1975). In some of the composing-aloud protocols we have collected, we can observe a writer reflecting on the accuracy of a particular phrase or debating the wisdom of including a given bit of information. Here, for example, is a transcript of a case worker composing aloud while writing a report of her visit with a child in a foster home:


I found him watching his shows on television. He is a delightful five-year-old child with blond...

Interrupts/ Speaks

Not, it’s not blond, it’s sort of dirty blond. Well, let’s see...


He is a delightful five-year-old child. Long lashes, cute, tall slender...


That’s what I wrote about him. Tall and slender.


He is a delightful five-year-old with long [eye] lashes


...and very intellectual looking?


...intellectual looking in his corrective lenses, in his glasses.


He is a delightful five-year-old with long lashes and very intellectual looking in his glasses.


That doesn’t sound right. No, cross all that out.


He is a delightful five-year-old.

Another example of composing aloud comes from a protocol in which an administrator was writing a memo to advise lower level supervisors that their workers were not following the correct procedure for filling out a particular form. The draft this administrator wrote during her composing-aloud session began with a request:

Please advise all workers that Form 189-B... is somewhat confusing to workers. This form should be used only when...

Apparently this request was very important to the administrator. She revised it considerably for her final draft. And almost half of the comments in her protocol concern this request. Although she was writing this request, she made the following comments, over a period of several minutes:

No, ah, let’s see. How do we word this one so that they don’t get uptight? Can’t demand, can’t ask, just.... Ah, let’s see. Do we advise them of the incorrect usage? Ah, oh dear. Writes: “Please remind....” Oh, Lord, how to be tactful? Let’s see, what do I want to do? I want to tell them that they... why they use form 189-B and that they’re doing it wrong.

In these comments, the administrator mentions concerns that also appear in the interviews we had conducted over a year earlier: How are my readers likely to react to what I am going to say? How can I create a persona that will cause as little undue stress as possible?

Our first example of composing aloud suggests that this procedure elicits certain types of information that cannot be obtained through the interview procedure we have described. The second example suggests that, for some writers, composing aloud elicits some of the same types of comments we encountered in our interviews. Having acknowledged these values of the composing-aloud procedure, we want to suggest some of its limitations. The most obvious limitation is that not everyone feels comfortable composing aloud. In one study, Cooper and Odell (1976) tried to get professional writers to compose aloud. Of the eight writers involved, only one gave a detailed report of the thoughts, feelings, and questions that attended his effort to write a draft. Most of the other writers simply read aloud the words they wrote on the page, responding briefly or not at all to our requests to “tell us what you are thinking as you write.” With workers in the department of social services, results were somewhat more encouraging. All four provided at least some useful information while composing aloud in response to this request, “Write a description of your job for readers of Seventeen magazine.” But when asked to compose aloud while doing her normal day-to-day writing tasks, one writer did nothing more than write and read aloud as she wrote. As we have noted, other writers provided somewhat more information. But when compared to their comments on their reasons for choosing one alternative in preference to another, this information seems relatively limited. Consider the writing of the administrator who was writing the memo about the correct use of form 189-B. When we examined a collection of her memos and letters, we realized that her writings were likely to vary in several ways:

The way she addressed her reader.

Whether she included introductory, context-setting material at the beginning of her writing.

The way she referred to herself.

Whether she shifted level of abstraction to elaborate on a given statement.

The way she phrased a command or a request.

Whether she concluded a memo or letter with a phrase inviting further communication (e.g., “if you have any questions...”).

The way she signed her name.

In her composing-aloud protocol, this writer commented on one of these variations—the way she wanted to phrase a request. Her comments touch on issues that also appear in her interviews. However, during her composing aloud, she makes no comment at all on other types of choices (listed above), even though our interviews led us to believe that these choices were not trivial for her. This administrator was the person, cited earlier, who had very definite notions about the usefulness of signing her name M. Smith, Margaret Smith, or Meg. Interestingly, the memo about form 189-B was signed M. Smith. Yet, the composing-aloud protocol did not contain any reference to a matter that, as we have seen, is quite important to this writer. It seems, then, that this writer constructed through experience a plan based on a quite complex knowledge of possible reader biases. Once that plan was formulated, the writer did not need to reconstruct it for each writing and thus did not need to attend to it or verbalize it. That a decision of this sort may not be made consciously for each piece does not diminish its importance, but it does suggest why it will not likely be revealed in a composing aloud protocol, which is more suitable for eliciting information about global processes, not about the specific knowledge and plans applied to familiar tasks.

We cannot make too much of this single omission from one composing-aloud session. Yet, the problem we have described consistently occurs in the composing aloud of the two caseworkers. From each of these caseworkers, we have three transcripts of their composing aloud while writing reports of their meetings with clients. None of these protocols contains any comment on types of choices that appear in the writing of every worker in their unit. Given this fact, one might wonder whether these choices are, in fact, as important as we have suggested. After all, if a choice were really important, would not a writer comment on it during the composing-aloud process? We agree, of course, that points mentioned when a writer composes aloud are worth our attention. When a writer deliberates over, say, a phrase, it seems reasonable to assume that the phrase may represent a significant rhetorical decision. But it does not seem reasonable to assume that composing aloud will enable a writer to comment on all the important choices he or she must make. As we know, the composing process is complex; in writing a sentence, a writer has to decide on a number of matters, ranging from the syntactic form in which a proposition may be cast to the appropriateness of expressing that proposition to the intended audience. Given this range of decisions, many of which must be made almost simultaneously, and given the limitations of STM, it is surely inevitable that a written text will entail significant decisions that cannot be remarked upon when one composes aloud.

In suggesting that composing-aloud protocols might omit important information, we raise a criticism that may be made of any research methodology, ours included. It is unlikely that a single methodology—in effect, a single perspective—will ever tell us all we need to know. Consequently, we think researchers should look for ways that several existing methodologies might be brought to bear on the same topic. For example, we think composing-aloud protocols might complement the information derived from our interviews. Composing-aloud protocols may be useful, for instance, in differentiating between that to which experienced versus inexperienced workers consciously attend when they write. We speculate that inexperienced workers—because they do not yet have the same knowledge of the rhetorical context and the way to manipulate language to achieve their purposes within that context—would have to devote more attention to constructing that knowledge when they write each piece. The composing-aloud protocols should reflect this difference and also provide information about how inexperienced workers build that knowledge. Composing-aloud protocols might also be a source of information about the strategies a writer uses to solve the unique problem presented by each writing task, more specifically the way context-specific knowledge is combined with more global-writing strategies to solve these problems. Such information about what writers know and how they use this knowledge, information derived from discourse-based interviews, and composing-aloud protocols could serve as a useful heuristic, particularly for inexperienced workers.

In suggesting one way research strategies might complement each other, we are making this assumption: Researchers in our field need a repertoire of research strategies, a repertoire that includes interviews, composing aloud, analyses of written products, and videotaping writers while they are writing. Our goal in this chapter is to add to that repertoire.

Works Cited

Atlas, M. A. Assessing an audience: A study of expert-novice differences in writing (Technical Report No. 3, Document Design Project). Pittsburgh, Pa.: Carnegie­-Mellon University, 1979. Contract No. NIE-400-78-0043.

Booth, W. The rhetorical stance. In W. Winterowd (Ed.), Contemporary rhetoric: A conceptual background with readings. New York: Harcourt Brace Jovanovich, 1975.

Britton, J., Burgess, T., Martin, N., McLeod, A., & Rosen, H. The development of writing abilities (11-18). London: Macmillan, 1975.

Cazden, C. The neglected situation in child language research and education. In F. Williams (Ed.), Language and poverty. Chicago: Rand McNally, 1970.

Cooper, C. R., & Odell, L. Considerations of sound in the composing process of published writers. Research in the Teaching of English, 1976, 10, 103-115.

Crowhurst, M., & Piché, G. L. Audience and mode of discourse effects on syntactic complexity in writing at two grade levels. Research in the Teaching of English, 1979, 13, 101-110.

Flower, L. S., & Hayes, J. R. The cognition of discovery: Defining a rhetorical problem. College Composition and Communication, 1980, 31, 21-32.

Gibson, W. Persona. New York: Random House, 1969.

Goswami, D. Rhetorical occurrences in occupational writing. Paper presented at the annual meeting of the National Council of Teachers of English, Kansas City, November 1978.

Kinneavy, J. A theory of discourse. New York: W. W. Norton, 1980.

Knoblauch, C. H. Intentionality in the writing process: A case study. College Composition and Communication, 1980, 31, 153-159.

Labov, W. The study of nonstandard English. Champaign, Ill.: National Council of Teachers of English, 1970.

Lloyd-Jones, R. Primary trait scoring. In C. R. Cooper & L. Odell (Eds.). Evaluating writing: Describing, measuring, judging. Urbana, Ill.: National Council of Teachers of English, 1977.

Moffett, J. Teaching the universe of discourse. Boston: Houghton Mifflin, 1968.

Odell, L., & Goswami, D. Writing in a non-academic setting. Research in the Teaching of English, 1982, 16, 201-224.

Polanyi, M., & Prosch, H. Meaning. Chicago: University of Chicago Press, 1975.

Rubin, D. L., & Piché, G. L. Development in syntactic and strategic aspects of audience adaptation skills in written persuasive communication. Research in the Teaching of English, 1979, 13, 293-316.

Scribner, S., & Cole, M. Literacy without schooling: Testing for intellectual effects. Harvard Educational Review, 1978, 48, 448-461.

Smith, E. R., & Miller, F. D. Theoretical note: Limits on perception of cognitive processes: A reply to Nisbett and Wilson. Psychological Review, 1978, 85, 355-361.

Van Dyck, B. Partial taxonomy of writing demands on bank executives. Paper presented at the annual conference on College Composition and Communication, Washington, D.C., March 1980.

Return to Composition Forum 49 table of contents.