Institutional discourse

MICHAELAGAR*

Abstract

Several studies of institutionell discourse are reviewed - discourse where one

person who represents an Institution encounters another person seeking its
Services. Focusing particularly on studies in the courtroom and the clinic, a
framework is developed that divides the discourse into diagnoses, directives
and reports. The framework not only organizes the different studies, but also
provides links to discursively based social theones such äs those ofFoucault
and Habermas.

l. Introduction

Discourse analysis takes äs its data any Stretch of talk or text. No constraints
are placed on the contexts within which the talk/text of interest was pro-
duced. One favorite type of discourse, for example, is called 'natural conver-
sation', implying a class of contexts that involve symmetrical social relations,
unconstrained topic flow, and infbrmality of style.
Another type—represented by a list of studies that is growing exponen-
tially-rests on discourse produced in contexts of quite another sort. One
person-a citizen of a modern nation/state—comes into contact with another
—a representative of one of its institutions. The discourse that results is some-
thing more than 'unnatural non-conversation'. I would like to take a sample
of studies of discourse that fit this loose definition of context-type and
examine them for pattern.
The proliferation of such studies in the past few years is remarkable.
Gumperz, one of the pioneers, deals with employment Interviews and court

0165-4888/85/0005-0147 $2.00 Text 5(3) (1985), p.p. 247-168

© Mouton Publishers, Amsterdam Brought to you by | University of Arizona
Authenticated
Download Date | 6/2/15 9:17 AM
148 Michael Agar

testimony (1982a; 1982b). Labov and Fanshel's approach to discourse grew

from an examination of a psychiatric encounter (1977). Erickson and
Shultz's pioneering work on 'gatekeepers' (1982) examines cases where
school counselors advise students. A recent special issue of Text focuses
exclusively on discourse analysis in courtroom settings (Danet, 1984). The
Center for Applied Linguistics issued a special volume on doctor/patient
communication (Fisher and Todd, 1983), which in turn was followed by
a special issue of Discourse Processes (Frankel, 1984). Other studies have
looked at discourse in federal government agency settings (Bilmes, 1981;
Agar, 1983). The work on discourse in educational settings between teachers
and students is too elaborate to even begin to review here (but see Shuy's,
1984, overview of medical, legal and educational discourse research).
Though all these studies fit our minimal defmition of 'citizen/institution'
context, the styles of discourse analysis vary. The discourse devices high-
lighted ränge from paralanguage and poetics to adjacency pairs and topic
shifts to speech acts and Schemata. In spite of their analytic diversity, a
reading of some of the specific studies convinces me that a general frame-
work is waiting to be made evident. It is important to develop it; it represents
an intermediate level between the human contact that led to the discourse
and the broader characteristics of the society in which it was embedded. The
literature review on which the framework is built is hardly exhaustive, but it
covers enough of the varied approaches to investigate what the different
analyses have to do with each other. To reduce the review to manageable pro-
portions, the studies of the courtroom and the clinic reported in the Text
issue and the CAL monograph are emphasized.

2. A sketch of the framework

My purpose is to review some discourse analytic studies in the courtroom and

the clinic to look for common patterns that characterize institutional dis-
course. The review will not deal with a topic that several touch on—the simple
problem of access to the same codes. The problem appears in the studies in
two ways. First, law and medicine especially have a Jargon all their own that
serves äs a barrier to the dient (Cicourel, 1981; O'Barr, 1981; Shuy, 1984).
Second, cultural, ethnic and class differences introduce further communica-
tive complications. (Gumperz, 1982; Wodak-Engel, 1984). Important äs
those problems are, they are by-passed in this article. We will presuppose
competence in the relevant codes and take it from there.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 149

The institutional discourse must accomplish three things. First, the institu-
tional representative must diagnose the client. Who is the client? Why is he/
she now in contact with the Institution? The Institution provides a limited
number of ways to describe people, their problems and the possible Solutions.
These ways are called Institutional Frames. Clients, on the other hand, come
to the encounter with a variety of ways of thinking about themselves, their
problems, and the institution's relationship to them. They have their own
Client Frames. Diagnosis is that part of the discourse where the institutional
representative flts the client's ways of talking about the encounter to ways
that fit the institution's. In our symbolic shorthand, diagnosis is the process
through which the institutional representative fits the client frame to the
institutional frame.
Another part of institutional discourse is the directives. They are one of
the goals of the diagnosis; the institutional representative directs the client to
do certain things or directs an organization to do certain things to or for the
client. A third part of the institutional discourse is the report. A report is the
summary of the institutional discourse that the institutional representative
produces. The client may not be present; in fact, the report, in written or oral
form, may be directed only to other institutional representatives. The institu-
tional frames prescribe how a report should look and what it should contain.
The segments of institutional discourse, then, are diagnoses, directives, and
reports. Actual cases of institutional discourse do not always follow so neat
a paradigm—a physician, for example, miglit ask a question (diagnosis), teil
the patient to do something (directive), and jot a note into the medical record
(report). Or he/she might ask a long series of questions, give the patient a long
lecture, and then go to his/her office to write a lengthy summary. The
concepts developed here are meant to organize a general understanding of
institutional discourse, not to predict the actual sequences of utterances of
which they are made.

2.1 Diagnosis

Several studies of institutional discourse were'reviewed for this article. They

share a contextual focus—contact between representatives of the institutions
of medicine or law and clients who come to those institutions because they
need to or are required to—the clients suffer from a physical or social path-
ology that the institutions are designed to repair.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
150 Michael Agar

No matter what the biases of the investigator, questions and answers play
a central analytic role. With few exceptiom, the institutional representative
does the asking, and the dient answers. Several of the studies present actual
frequency counts to document this claim (Todd, 1983, and Philips, 1984,
for example). West (1983) tallied 773 questions in 532 pages of transcript.
91% were initiated by the institutional representative ( a doctor, in this case);
9% by the client-patients.
The goal of diagnosis requires the institutional representative to fit the
client's problem to the institution's frames; therefore, the active questioning
of the institutional representative is not surprising. But the studies go beyond
a description of a neutral elicitation of information. Instead, reading across
the different studies, the institutional representative Starts to look heavy-
handed. He/she takes a discourse device—the initial turn in a q/a adjacency
pair—uses it to drag the client, willing or not, through the institutional repre-
sentative's topic of choice. When the client does ask questions, it is often at
the invitation of the institutional representative, besides, äs West found out,
half the client questions were 'marked by speech disturbances' (1983:100),
signalling their own recognition that a client question is out of character.
If we shift from the clinic to the courtroom, we find the same institutional
representative control described:
First, the whole dialogue is organized along question-and-answer lines with
the judge asking the questions (with one notable exception) and the witness
providing the answers. This confirms the results of previous studies on court
interaction in different contexts demonstrating the fixed question-answer
format of court interactions. Moreover, the judge's questions sharply limit the
ränge of alternatives open t o the witness. Some of his questions, particularly
in the opening sequence, are Statements intoned äs questions requiring
consent. Later, when addressing new points, the judge often begins with a
series of queries starting with an open question which is then narrowed to
focus on a particular point by the more specific questions that follow. Thus,
the judges shape the hearing into an 'interrogation' from the start. While this
is an economical procedure, it clearly violates the procedural law according to
which the witness should be requested, at the beginning of the hearing to
provide a coherent description of the events in question "in his own words"
(Ceasar-Wolf, 1984:202, citations omitted).

The quantitative count and the qualitative quote both report a conclusion
about a particular discourse device and its institutional discourse function
that appear in study after study. Questions are used by the institutional repre-
sentative to control the flow of information. Shift the perspective from

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 151

questions to the topics they introduce, and we get the same conclusion. Shuy
(1983), known for his topic analyses of medical and legal institutional dis-
course, concludes that topic introductions and reintroductions are typically
the institutional representative's perogative.
The global conclusion—that institutional representatives use questions
both to elicit Information and to control institutional discourse—is robust.
But when we lift the top off the 'question' concept we find a variety of inter-
pretations about what kind of utterances count äs what kind of questions.
The simple idea of the q/a adjacency pair is actually far from simple. There is,
first of all, the traditional ethnomethodology discussion that introduces such
complications äs insertion and side sequences. Then the elaborate work of
Labov and Fanshel (1977) shows the many 'indirecf ways that requests are
posed in a psychiatric interview. Even when we move onto the solid ground
of direct Speech acts, an immediate distinction between yes/no questions
and wh-questions appears. Wh-questions are generally thought to be less coer-
cive, but in some circumstances Philips suggests they may be less polite
instead (1984). She also further complicates the issue by distinguishing four
types of yes/no questions—the usual form, a declarative form, a deleted aux
form, and an anaphoric tag.
West turns an ethnomethodological eye to the problem and shows that
q/a pairs differ in institutional discourse function. In her quantitative work—
mentioned above—she only counted q/a pairs where the question pointed
towards the answer. She did not count questions that pointed backwards—
'requests for repair of a previous item, requests for confirmation of a previous
item, and markers of surprise at a previous item' (1983:85). The q/a pairs she
rules out in fact represent a crucial area of client participation, since, äs Shuy
notes, the client can only '. . . request clarification, Interrupt, pause, express
hesitation or uncertainty even in the presentation of a response, agree, or
respond directly' (198 3:196).
Another slant on the function of q/a pairs is that some institutional repre-
sentatives provide a question that allows the client to demonstrate his/her
competence. In fact, this 'competence slof was critical in Fisher's study;she
notices that the clinic catering to higher SES womea more frequently
provides such slots, and that their presence, together with a competent
answer, is a condition present in the relatively few cases of institutional dis-
course where the client influenced the treatment decision outcome (Fisher,
1983).
Another interesting fly in the q/a ointment is that students of educational

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
152 Michael Agar

institutional discourse, a field not covered in this article, point out that
'adjacency pair' doesn't cover the international territory. Instead, teacher/
Student exchanges have a third part, the teacher's evaluation of the
preceding teacher question and Student answer (Mehan, 1979; Stubbs, 1983).
With the help of D'Andrade's speech-act coding System, Todd looks at
some doctor/patient talk and finds a similar structure there. Like other
students of institutional discourse, she points out that the institutional
representative asks the questions and the client provides the answers. But
then the institutional representative—the doctor—'reacts'.

The doctor's reactive serves two purposes. The doctor initiates the interaction
with a question. In so doing, the doctor also controls the topics discussed in
their relation to the patient's presentation. The reactive serves, first, to end
the interactional segment and the topic, and second, to bring control of the
interaction back to the doctor, allowing the doctor to end that frame and to
initiate a new one (1983:165).

Not only does the institutional representative control through question and
topic; he/she also metacomments on the adequacy of the client's responses.
Todd (and the educational researchers) describes metacomments in a struc-
tured three-part sequence, but, by extension, we can expect metacomments
by the institutional representative to appear in other, less structured ways äs
well.
No matter how you look at it, the overwhelming conclusion from the
court-room and clinic studies is that the institutional representative asks the
questions and the client provides the answers. Questions and meta-evaluations
of the answers, control of the topic flow, and determination of the territory
within which the client can talk are in the hands of the institutional represen-
tative. In the verbal dance of institutional discourse diagnosis, the institution-
al representative takes the lead.
The studies emphasize questions and topic shifts; but scattered through
the different cases are other devices äs well, though these are less consistently
used. Different authors report other ways to gain control—such äs back-
channelling, Interruption, the melodic, rhythmic, poetic and prosodic
qualities of language, deictic shifts, forms of address, lexical selection, register
—in fact, this list is not exhaustive. Many devices are candidates for the
diagnosis control function, but for the present, we will emphasize the q/a
and topic shift issues, just äs is emphasized in the various articles.
The model of institutional discourse diagnosis shows the institutional

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 153

repräsentative steering the discourse within a particular topic framework

(Brown and Yule, 1983). With these devices at his/her command, the institu-
tional representative uses his/her control to fit the client into the organiza-
tional ways of thinking about the problem. This knowledge—the institutional
frames—is the source of the questions that the institutional representative
poses. He/she is trying to determine which frames are relevant and then
having made that determination, to fill in the variable slots with appropriate
Information from the client. The client, on the other hand, has his/her own
frames—the client frames. The client frames organize the problem in different
ways. The institutional representative uses the devices to select and fill in the
institutional frames, no matter what client frames are presented.
Several articles deal with the lack of fit between the institutional frame
and the client frame. Consider the doctor/patient example. At first, äs we
expect, the doctor controls the topic and ränge of answers with his questions.
Then he asks for the patient's Interpretation. The patient makes eye contact,
gives longer answers, and discusses his problem with social security, long-
term disability and employment. The doctor then:

returns the interview to his domain of expertise-analysis and treatment of

the patient's biomedical Problems . . . He does not comment on or assess the
problem identified by the patient (loss of income), nor does he at this point
or at any other point in the discussion offer a remedy. The question thus
raised is one of the negotiation and Status of patient Problems and the
selective attention paid by the provider to the biomedical dimensions of care
(Treichler et al., 1984:68).

Other analyses also detail the institutional frame/client frame relationship.

Page t (1983) shows a physician using discourse devices to steer away from a
woman's concern with the personal and social consequences of cancer into a
medical frame dealing with gen^ral health and nervousness. Cicourel (1983)
describes how the physician uses an institutional frame to Interpret a client's
behavior äs 'bereavemenf when, in fact, according to her client frame, she is
anxious that she 'caught' cancer from her deceased hasband. In a courtroom
example, Ceasar-Wolf (1984) describes the 'discrepant relevancies' between
the institutional frame and the client frame. The judge must take the 'relevan-
cies' in a story from the witness's point of view and shape them into a story
with the proper 'relevancies' for the court. In another example, Liebes-
Plesner (1984) introduces the twist of two institutional representatives with
competing diagnoses. Two lawyers in a rape trial in Israel use discourse

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
154 Michael Agar

devices to control the client into two distinct and opposing Images—girl vs.
woman, vulnerable vs. scheming, childish innocence vs. sexy and street smart,
and so on.
After you read through the cases, you begin to wonder if things could be
any other way. There are cases of client questions, sometimes ignored, some-
times well and competently answered, but they are rare. The best Vignette is
reported by Treichler and her colleagues. After the institutional representa-
tive (a doctor) finishes diagnosing a patient, he says that 'if you don't mind' a
Student Vould like to maybe talk with you before you split, ok?' the shift to
polite and mitigating discourse devices is striking. The Student says Tm just
kinda curious um about the pain you're having now'. The Student is silent,
allows pauses, and encourages the client to develop his story. He learns that
client sees a psychiatrist, has completed an alcohol program, and takes
thorazine. None of this is taken up by the resident, nor does it enter into the
record. The Student is not an institutional representative, and the talk was not
an official part of the institutional discourse.

2.2 Report

The record that the doctor just ignored in the clinic example is a critical part
of institutional discourse. Diagnosis, once successfully accomplished from the
institutional representative's point of view, is a means to an end. One end is
the report that goes into the archives. Among the studies reviewed here,
several investigate the relationship between the verbal discourse and the re-
port that it generates. Cicourel (1983), in fact, views the relationship between
institutional discourse and reports äs being the core problem. His analysis
focuses on the interpretive processes between medical diagnoses and medical
reports. In his case, the concern comes äs no surprise. Cicourel is part of the
ethnomethodological tradition that developed around the relationships
between organizational records and the human interaction that led to them.
Shuy looked at three case-history Interviews, also in a medical setting, and
described this complicated relationship äs well.

It is not surprising that in a medical interview that clearly follows a visible

written questionnaire format, there is a rather rigid topic introduction
sequence. However rigid that format may be, one startling conclusion faced
me at the end of my examination of some 100 Interviews: It would be very

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 155

difficult to reconstruct the written questionnaire on the basis of the tape-

recorded Interviews (1983:194).

Fisher opens her article with an observation that women who go to one clinic
get hysterectomies more often than women who go to another. Nothing in
the medical records, she observes, explains why this should be so (1983).
Other case studies of institutional discourse also deal with reports. Some
focus on the characteristics of 'legalese', the nature of the report code that is
neglected in this article (Danet, 1984; Kurzon, 1984). Others examine the
form of verbal rather than written reports. In a case described earlier, two
competing institutional representative attorneys in a courtroom—used dis-
course devices to construct two different images of the same client. Ceasar-
Wolf (1984) shows how the inquiring judge works towards a 'plausible and
consistent' story to account for differences in client descriptions of a traffic
accident. Tannen and Wallat (1983) show how the institutional representative
describes a problem äs more serious when reporting to her colleagues,
compared with the aloof tone she used with the client.
Reports are critical; they may be the goal against which the progress of the
diagnosis is measured. In a general way, that is what the institutional repre-
sentative does when he/she fits the client frame to the institutional frame.
The move from diagnosis to report calls into play a new set of interpretive
procedures that organize the problem in an institutionally prescribed way.
The transformation from diagnosis to report may occur during the face-to-
face institutional discourse, or the institutional representative may produce it
solo. In whatever form, a report is another example of institutional represen-
tative control of the outcome of institutional discourse.

2.3 Directive

'Diagnosis' and 'report' are necessary constituents of institutional discourse.

They may also be sufficient. In fact, lack of directives might be a client
complaint. One goes to an institutional representative, has one's problem
(mis)diagnosed, a report is filed, and that is the end of it. Or perhaps
directives are not necessary. A client consults a doctor or an attorney, diag-
nosis occurs, a report is filed, but directives are not required.
On the other hand, directives may be a regulär part of iristitutional dis-
course. Like other researchers, Todd (1983) notices that doctors issue direc-

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
156 Michael Agar

tives while patients do not. That is natural enough, since the client goes to the
doctor seeking guidance on what to do about a problem, äs well äs advice on
what the problem is. Adversary lawyer institutional representatives in a court-
room also recommend directives to the judge, who in turn must select and
implement them. Later in this article, when a couple of other institutional
discourse cases are described, directives will also be seen to play a critical role.
But by and large, the studies reviewed here deal more with diagnosis, less so
with reports, and least of all with directives. Perhaps the study bias reflects
real institutional discourse priorities; the greatest amount of effort goes into
fitting the variety of client frames found in any society into a small set of
institutional frames. The next problem is to map the institutional frames onto
a still smaller set of report frames. At that point, the directive is almost
automatic. That is what institutions are supposed to provide—efficient ways
to handle routine problems. On the other hand, the neglect of directives may
come from a narrow focus on the verbal part of institutional discourse, a
natural focus for someone interested in discourse.

3. Discourse ecology

In several studies, the authors note that one reason for institutional discourse
control is a response to pressures about which the institutional representative
can do nothing. Ceasar-Wolf, in the elaborate quote cited earlier, said that the
judge's control may be an 'economicaP procedure (1984). West (1983) talks
about time pressures on the institutional representative. Paget (1983) shows
how a physician uses discourse devices to steer the institutional discourse
away from the client frame of cancer. Yet she also describes how medical
readers of the transcript see the physician institutional representative's talk äs
being, in fact, about the patient's cancer. Perhaps the medical reader/listener
has adequate institutional frames to connect text to topic with inferential
chains while the client, lacking those same institutional frames, cannot do
so and therefore concludes that the cancer topic is fenced off by the institu-
tional representative's discourse devices.
Efficiency, economy, time pressure and background knowledge; there are
circumstances around the institutional discourse over which neither the
institutional representative nor the client have any control. We will call these
circumstances the discourse ecology of the institutional representative and
the client. The concept is critical, since it shifts our explanatory focus away

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutionell discourse 157

from individuals toward Systems. Why is the institutional repräsentative so

controlling? Consider two hypotheses about complex societies:
1. Institutional frames represent specialized knowledge. Their apprehension
requires training and experience. The increasing rate of Information
growth creates. proliferating institutional frames and more specialized
institutional representatives.
2. Though there are counter-examples, typically private-sector organizations
have äs a core goal the minimization of costs to maximize profits. Public-
sector organizations must often maximize inadequate resources. In either
case an institutional representative is stressed continuously to make do
with less.
If these hypotheses hold, then an institutional representative who wants to
hand control over to the client could not afford to do so. First, the institu-
tional representative has to explain the institutional frames; the explanation
might involve some teaching, depending on how elaborate and technical the
institutional frames are. Even after the teaching, the time required to develop
and understand the client frames in collaboration with the client would be
substantial. Where the client frame was simply wrong (like 'catching' cancer),
the institutional representative would also explain why that part of the
client frame was in error. Given the ecology of the institutional representa-
tive, the time required for such privileged treatment of the client would
consume the private profits or public budgets, and the institutional represen-
tative would be out of a Job. Just to complicate things, add the possibility
of a recent institutional representative history of unemployment in a glutted
labor market. The ecology now contains an additional incentive for the effec-
tive use of discourse devices to fit the client frame to the institutional frame
äs quickly äs possible.
Not that all institutional representatives handle institutional discourse in
the same way. Far from it. Even in the few studies sampled for this article,
institutional representatives appear in several lights, ranging from condescen-
ding and authoritarian to concerned and open. Any reader of this article has
him/herself experienced a variety of institutional representative types—the
'goes by the book' institutional representative who clings to the offically pre-
scribed institutional frames, the 'nasty' institutional representative who uses
his/her position to browbeat clients, the 'reasonable' institutional represent-
ative who works more loosely to Interpret client frames to the institutional
frames, the 'advocate' institutional representative who tries to map the client
frame onto the institutional frame that will provide the best outcome for the

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
158 Michael Agar

client. Likewise, anyone who has been in an institutional repräsentative role

(which also includes most readers of this article) knows there are different
types of client. But whatever the personal styles, the discourse ecology
sets limits around what either the institutional representative or the client can
can do during institutional discourse.

4. Two informal cases

In this article I have used cases from the courtroom and the clinic. A huge
literature on discourse in educational settings also exists. In addition, other
discourse studies cover institutional discourse in occupational settings. Just to
foreshadow some of the complications that any institutional discourse frame-
work must eventually handle, I would like to describe two additional cases
from my own experience. The first case is a hearing held by the Interstate
Commerce Commission a few years ago. (The study is published elsewhere
Agar, 1983). Briefly, the hearing dealt with an issue about a 'surcharge'
paid to independent truckers to compensate them for the rising cost of fuel.
The system that had been in force was unsatisfactory, said the ICC, so a
public hearing was held to hear the views of different participants in the
trucking industry on which alternative system would be better.
Notice the differences. The institutional discourse begins with a series of
proposed directives rather than a diagnosis. The discourse ecology immedi-
ately comes into play, since so many clients wanted to testify that each was
only allowed a few minutes. Each client gave a report and a directive. The
institutional representative—in this case a panel of ICC commissioners— then
moved into the diagnosis, asking questions to map the client frameworks just
presented to the institutional frameworks that they were using to organize
their decision. The institutional representatives then retired from the open
hearing and at some later time met, discussed the case, and eventually
published a report of their decision. That decision, in turn, was taken to court
by those clients upon whom it had a negative impact, leading to another
series of institutional discourses of the type examined in the court studies.
The framework fits, but there are some differences:

1. Rather than dealing with an isolated case, we see a linked series of

institutional discourses running across several organizational domains.
Hearings were held all over the country, which led to the main public

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 159

hearing in Washington, which led to a public document, which led to

court cases, and so on. Such linkage is acknowledged in some of the
studies reviewed here, but this example shows how the study of some
institutional discourses will have substantial organizational Span and
historical depth.
2. The clients had a chance—however brief-to take the floor and present
their frameworks. In fact, it was the analysis of the common client
frameworks underlying the reports of the independent trucker witnesses
that was the goal of my article. Each client also produced a directive. In
the early part of this article, the point of view of the institutional
representative was emphasized. A füll framework will have to allow for
different points of view on the institutional discourse; there may also be
a client diagnosis, a client directive, a client report, and a client discourse
ecology. It is not surprising to find room for the client point of view in a
public-sector hearing in a society chartered äs a democracy. It is also not
surprising, given what we have learned in the microanalyses of institu-
tional discourse reviewed earlier, that the commissioners decided on a
closed meeting and then the problem moved into the courts where
institutional representatives could take charge.

The next case I would like to informally describe is the IRS office audit. I
have been through four of them in about ten years and consider myself, if
not an analyst, at least an experienced client. In this case, control is estab-
lished at the outset because the client is required to appear. The diagnosis
proceeds pretty much äs described earlier, with one notable exception. If the
client introduces a topic, the institutional representative will sometimes allow
it to see what additional Information they can learn about the client relevant
to the audit. If no such helpful Information is forthcoming, the institutional
representative uses discourse devices to regain control and put the audit back
in Service of the institutional frames.
Clients also learn the institutional frames with experience. Knowing them
allows the client to 'pre-package' Information to fit those institutional frames
to his advantage. After the audit, the client receives a copy of a written report
that also goes into the archives. The report contains directives to the client
and to the agency. If you hadn't been there, you wouldn't know why they
were made. The client can then request a review by the auditor's superior—a
kind of 'meta-institutional discourse'. After that, the client must go to the
courts.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
160 Michael Agar

The IRS audit fits the framework a little more clearly, partly because the
face-to-face discourse between a single institutional repräsentative and a single
client is the canonical form described in the articles. But again, there are some
noteworthy differences.
1. The client does not come to the organization for help with a problem.
The organization .requires him/her to appear with sanctions looming over
his/her head if the invitation is ignored.
2. The client can introduce topics during diagnosis, but only because it is in
the institutional representative's interest to allow it, a more malevolent
version of the 'competence display slot' provided by the doctor des-
cribed earlier.
3. Institutional discourses are linked again, only this time a hierarchical
linkage is emphasized—an appeal to the auditor's superior is an institu-
tional discourse about an institutional discourse.
I use these two informal examples—one from research and one from
personal experience—to make two points: (1) The framework for under-
standing institutional discourse resulting from the studies of courtroom and
clinic has a broader application than just the cases that generated them;T)ut,
(2) äs the framework is applied across wider ranges of institutional discourse
types it will have to be enriched to accommodate them. The point of this
article is t o suggest a framework, not conclude it.

5. Toward some theoretical Integration

In this article discourse analysis has been used äs a 'theory of the data', a
series of guidelines to untangle multi-level patterns in a complex structured
object. The question now is, data in service of what broader issues? We have
some of the raw material of citizens and institutions in contact. If we look at
a number of cases, what kinds of inferences can we make about the society
from which they were taken? We need to exchange the microscope for a
wide-angle camera.
Consider, for example, Foucault's key discourse hypothesis:

that in every society the production of discourse is at once controlled,

selected, organized and redistributed by a certain number of procedures
whose role is to ward off its powers and dangers, to gain mastery over its
chance events, to evade its ponderous, formidable materiality (1984:109).

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutionell discourse 161

After the discussion of institutional discourse, that should sound familiär.

Foucault, describing epochs in terms of the regulated distribution of power
and knowledge, recognizes that discourse is the expressive trace of those
distributions, the key raw material for the investigation of knowledge and its
controls.
Foucault is not the only one to recognize discourse äs an expression of the
social order. Consider the work of Habermas on 'universal pragmatics'.

In the final analysis, the Speaker can illocutionarily influence the hearer and
vice versa, because speech-act-typical commitments are connected with
cognitively testable validity Claims—that is, because the reciprocal bonds have
a rational basis (l979:63).

There are four validity Claims. Comprehensibility is the first. The other three
are (1) that the Speaker utter a true proposition; (2) that he/she wants to be
truthful or trustworthy; and (3) that a normative background obtains within
which the utterance is appropriate. The validity claims, then, lead to three
'felicity conditions'—the Obligation to provide grounds, the Obligation to
prove trustworthy, and the Obligation to provide justification.
I cannot ränge comfortably through the prolifk and often difficult
writings of Foucault and Habermas, and do not mean to imply that tlieir
theories are the only two candidates for placing detailed discourse analyses of
single cases in broader perspective. Nor do I think that we (the nose-to-the-
data discourse analysts) need them more than they (the grand theorists
unhindered by the complex and intricate details of real-world social action)
need us; good arguments can be made to the contrary. For the present, I only
wish to experiment with the institutional discourse framework äs a mediator
between grand theory and case study.
Foucault develops a series of 'rules' to focus on how discourse control
is accomplished—historically developed and institutionally supported rules of
exclusion, rules of limitation where discourse exercises its own control, and
rules that determine the conditions for discourse use. His rules of exclusion—
what you can't say, what's crazy to say, what's useless or unprovable—
certainly ring true äs three characteristics of institutional framework bound-
aries. For medlcine and law, his 'disciplinary' rule of limitation also applies;
before a proposition is eligible for expression it must already be 'within the
true', and his 'commentary' limitation rule, where a particular narrative is
simply the token of a Standard type, models some of the discussion of reports.
Finally, his rules on conditions for discourse deal with 'fellowships of dis-

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
162 Michael Agar

course' and 'doctrinal groups' in ways that suggest a social coherence to the
institutional repräsentative role. But the detailed analysis of single cases blurs
some of the distinctions that look so crisp in the social-theoretic air. The
institutional framework boundaries are not razor-sharp, and some of institu-
tional discourse, in spite of the institutional representative's controlling role,
is a mediation of the client framework and the institutional framework. In
the medical/legal cases, the discipline, the fellowship and the doctrinal group
may be three redundant labels for the same phenomenon. Foucault could use
some data.
Habermas offers a complementary view of the case studies, though there is
an initial complication. His universal pragmatics doesn't apply to 'institution-
ally bound' speech actions. As he himself says:

To be sure, "institutional bond" is a criterion that does not always permit an

unambiguous classification. Commands can exist wherever relations of
authority are institutionalized. (1979:39)

Our first gift to Habermas is an amplification of this complication. Is any

discourse whose felicity conditions are grounded in an institutional frame-
work 'bound' and therefore ineligible for validity-claim analysis? If institu-
tional discourse is ineligible, a significant portion of modern life is mied out
of the analysis. Habermas would not so restrict the ränge of his theory. For
now, we will pass through the problem and use his universal pragmatics on
the institutional discourse data.
Universal pragmatics brings questions of legitimacy into the analysis. Our
first observation is that institutional discourse is organized t o prevent raising
validity claims. Pressures in the discourse ecology and the discourse control
of the institutional representative leave a claim-laden client without much
chance to utter them. One client Option, if he/she were truly dissatisfied,
would be to bail out of the institutional discourse; but because of coercion
or needs, that is seldom a practical possiblity. Another Option is for the client
to manipulate the organization through the institutional representative, using
a knowledge of the institutional frames, an ability to read and manoeuvre the
institutional representative's style, and discourse devices such äs the 'backward
pointing' questions that West ruled out in her study. On the other hand, some
institutional discourse studies (Tannen and Wallat, 1983; Fisher, 1983)
describe institutional representatives who make room for validity claims,
though the claims are only possible because the institutional representative
allows it.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 163

If validity Claims are controlled by institutional discourse, how can they

ever be raised? One possibility is that validity Claims are raised on the level of
organizations rather than individuals. Political lobbies like the consumer,
environmental, and fundamentalist religous movements are examples of client
-oriented organizations that carry on validity-claim-based discourse with
private- and public-sector organizations. Validity-claim discourse is conducted
at an inter-organizational level. Although the media sometimes reports on
cases of the successful client versus the large organization, the cases are news-
worthy just because they are so exceptional. In modern society, the lone
client can't do much. He/she needs another organization to represent hirri/her.
Institutional representatives have their own problems raising validity Claims.
'Whistle-blowers' violate the sociability of their 'doctrinal groups' and 'fellow-
ships of discourse'. Though they may be canonized by the public, they
typically lose their institutional representative role and become institutional
pariahs.
Habermas' universal pragmatics highlights institutional representative
control in an interesting way; institutional discourse prevents client vaüdity
Claims, unless the institutional representative makes room for them. The dis-
course ecology suggests that it is unrealistic to expect much treatment of
validity Claims at any rate. On the other hand, the institutional discourse
analysis shows that universal pragmatics might apply to inter-organizational
discourse, a type not treated in this article.
From Foucault, institutional discourse finds theoretical ties with the
global characteristics of the social order äs regulated distribution of power
and knowledge. From Habermas, it links to a critical theory that sets out äs
its goal a social order based on reasonable answers to Claims on the validity
of its practices. From a reverse angle, institutional discourse contributes a
grounding of such theory that complicates them and the ränge of their appli-
cation, while also raising practical issues concerning their prescriptions for the
nature of the 'good society'. This brief discussion is only a nod in the
direction of a dialogue between grand theory and case study, limited in both
the theories and cases considered, but it is a serious nod inviting further and
less superflcial dialogue. I hope it shows that institutional discourse, or some-
thing like it, is a necessary intermediate level of theoretical discourse in terms
of which the conversation can proceed. Between the case and the state lie
vacant intermediate levels in need of conceptual structures. No linguist would
analyze phonology using concepts from semantics; no Computer scientist
would use theories of integrated circuits to describe an expert system. Why

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
164 Michael Agar

should we expect simple translation between discursive practices and

discourse devices? Institutional discourse is a single pitch in the game of
fillingin the gap.

6. Toward some complications

Though the institutional discourse framework mediates between case and

theory in a way that links the social order to the discourse devices by which i t
is accomplished, a variety of problems remain. It is beyond the scope of an
article to elaborate on them adequately, but I would like to point some of
them out:
1. The framework focuses on the contact between a representative of an
Institution and a client. Yet in some studies the discourse among institu-
tional representatives is a factor äs well (Bilmes, 1981). Our focus has
been on how institutions and clients interact; another branch of the same
framework must deal with how the co-members of the Institution dis-
cursively make it work.
2. The client has been neglected, just äs he/she has been in most of the
studies reviewed here. Clients come with varying degrees of sophisti-
cation in the institutional frameworks, with wildly different client
frameworks, with different abilities to psych.out an Institution, and with
different personal styles and manipulative abilities. I worked for many
years with a group—urban American heroin addicts—who are expert
clients. The best among them can leave discourse control in the hands of
the institutional representative and still get what they want. Interper-
sonal manipulation— the classic strategy of the powerless—belongs in the
framework äs well.
3. The whole question of what an Institution is still dangles in the breeze.
In a way, the framework begins a definition something like 'a socially
legitimated expertise together with those persons authorized to imple-
ment it'. This definition would also include Computer programmers,
automobile mechanics, real-estate salespersons, traffic managers, and
Professors. Medicine and law are certainly two of the most powerful
institutions in any modern state. Future work with institutional dis-
course cannot limit itself to just those two, however. Different types of
institutional representatives may use different discourse devices. A
real-estate salesperson, for example, has to do more work figuring out the

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutional discourse 165

dient frameworks; he/she must also create the client's need to stay with
him/her. More variety in the cases will require enrichment of the Frame-
work.
The relationship of Institution to its organizational context also awaits
clarification. Most of the institutions reported on in this article—
medicine, law and government-have been examined in organizational
contexts—hospitals, courts and meeting rooms. But not all institutional
representatives will operate in organizational settings, at least not all of
the time. Further, some organizations are built on a single institution-
like a hospital—while others draw institutional representatives from
several different institutions—like a large corporation that is both verti-
cally and horizontally integrated.

7. Conclusions

We have come a long way from the q/a adjacency pairs that began the
dis;ussion. We are left with several propositions:
1. Contact between representatives of institutions and those clients who
need to or must deal with them generates institutional discourse.
2. Case studies of institutional discourse highlight discourse devices that
allow for institutional representative control. The devices include the
initial position in q/a pairs, topic shift, and evaluative metacomments in
diagnosis, äs well äs the directives issued and the report prepared by the
institutional representative.
3. Control is exercised to remain within topic frameworks motivated by
institutional frames.
4. From a broader perspective, the control maintains the boundaries of
current discursive practices and serves äs a barrier against validity Claims.
5. In any particular instance, the discourse ecology suggests that even a
motivated institutional representative/client pair can do little to alter the
institutional discourse. An institutional representative can be more
sympathetic and a client can be more manipulative, but that is about it.
6. Changes in institutional discourse can come about with a shift in dis-
course from institution-person to organization-organization. You cannot
fight city hall, unless of course you're a judge or a corporation executive
offlcer.
7. We need a better sense of what institutions are, viewed from a discourse

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
166 Michael Agar

analytic point of view, äs well äs a better sense of how institutions are

embedded in organizational contexts. We also need more cases of institu-
tional discourse that describe institutions besides law, medicine and
education.
There are some other problems. As mentioned in the introduction, we
have neglected the research on class and cultural differences in communica-
tion, though some studies show their importance (especially Wodak-Engel,
1984). We have focused on the verbal, although several studies deal with
videotapes (Frankel, 1983; Tannen and Wallat, 1983). The bare bones of
institutional discourse has been described, although Tannen and Wallat show
that other discourse work can go on äs well, especially if one client is a child,
another is its mother, and a training videotape is being made at the same time.
Even with these and other limits, problems and gaps, institutional discourse
serves äs a mid-level framework that organizes the results of discourse
analyses done in a variety of clinical and court settings. It does so in a way
that shows how institutions work on the ground, while at the same time
setting up empirical material for use in developing critical theories of the
social order. It, or something like it, can both broaden the implications of any
single case study while at the same time showing the consequences of social
theory for practical action.

Note

Although I have not dealt with them all, excellent comments on an earlier draft
were provided by Erve Chambers, Linda Coleman, George Dillon and Kit Woolard.
Conversations with Deborah Pruitt also contributed to the final shape of the article.

References

Agar, Michael (1983). Political talk: Thematic analysis of a policy argument. Policy
Studies Review 2:601-614.
Bilmes, Jack (1981). Proposition and confrontation in legal discussion. Semiotica 34:
251-275.
Bogoch, Byrna and Danet, Brenda (1984). Challenge and control in lawyer-client inter-
action: A case study in an Israeli legal aid office. In B. Danet (ed.), Text 4(1-3):
249-275.
Brown, Gillian and Yule, George (1983). Discourse Analysis. Cambridge: Cambridge
University Press.
Cicourel, Aaron V. (1981). Language and medicine. In Language in the USA, Charles A.
Ferguson and Shirley Brice Heath (eds.) Cambridge: Cambridge University Press.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
 Institutionell discourse 167

- (1983). Hearing is not believing: Language and the structure of belief in medical
communication. In The Social Organization of Doctor-Patient Communication.
S. Fisher and A.D. Todd (eds.). Washington: Center for Applied Linguistics.
Caesar-Wolf, Beatrice (1984). The construction of 'adjudicable' evidence in a West
German civil hearing. In Studies in Legal Discourse, B. Danet (ed.), Text 4(1-3):
193-224.
Danet, Brenda (1984). Studies of Legal Discourse. Special Issue of Text, 4(1-3).
- (1984). The magic flute: a prosodic analysis of binomial expressions in legal Hebrew.
In Studies in Legal Discourse, B. Danet (ed.), Text 4(1-3): 143-172.
Erickson, Frederick and Shultz Jeffrey, (1982). The Counselor äs Gatekeeper. New
York: Academic Press.
Fisher, Sue (1983). Doctor/patient talk: How treatment decisions are negotiated in
doctor-patient communication. In The Social Organization of Doctor-Patient Com-
munication, S. Fisher and A.D. Todd (eds.). Washington: Center for Applied Lin-
guistics.
- and Todd, Alexandra Dundas (eds.) (1983). The Social Organization of Doctor-Patient
Communication. Washington: Center for Applied Linguistics.
Foucault, Michel (1970). The order of discourse. In Language and Politics, Michael
Shapiro (ed.). Oxford: Basü Blackwell, 1984.
Frankel, Richard, M. (1983). The laying on of hands: Aspects of the Organization of
gaze, touch, and talk in a medical encounter. In The Social Organization of Doctor-
Patient Communication, S. Fisher and A.D. Todd (eds.). Washington: Center for
Applied Linguistics.
- (ed.) (1984). Physicians and patients in social interaction. Special Issue of Discourse
Processes 7(2).
Gumperz, John J. (19.82a). Discourse Strategies. Cambridge: Cambridge Univers^ty Press.
- (I982b). Fact and inference in courtroom testimony. In Language and Social Identity,
John J. Gumperz (ed.). Cambridge: Cambridge University Press.
Habermas, Jürgen (1976). What is universal pragmatics? In Communication and the
Evolution of Society, Thomas McCarthy (tr.). Boston: Beacon Press, 1979.
Kurzon, Dennis. (1984). Themes, hyperthemes and the discourse structure of British
legal texts. In Studies in Legal Discourse, B. Danet (ed.), Text 4(1-3): 31-56.
Labov, William and Fanshel, David (1977). Therapeutic Discourse. New York: Academic
Press.
Liebes-Plesner, Tamar (1984). Rhetoric in the service of justice: The sociolinguistic
construction of stereotypes in an Israeli rape trial. In Studies in Legal Discourse,
B. Danet (ed.), Text 4(1-3): 173-192.
Mehan, Hugh (1979). Learning Lessons. The Social Organization ofClassroom Behavior.
Cambridge: Havard University Press.
- (1983). The role of language and the language of role. Institutional Decision-Making.
Language in Society 12:187-211.
O'Barr, William B. (1981). The language of the law. In Language in the USA, Charles
A. Ferguson and Shirley Brice Heath (eds.). Cambridge: Cambridge University Press.
Paget, Marianne (1983). On the work of talk: Studies in misunderstandings. In The
Social Organization of Doctor-Patient Communication, S. Fisher and A.D. Todd
(eds. ). Washington: Center for Applied Linguistics.
Philips, Susan U. (1984). The social organizations of questions and answers in courtroom
discourse: A study of changes of plea in an Arizona Court. In Studies in Legal Dis-
course, B. Danet (ed.), Text 4(1-3): 225-248.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM
168 Michael Agar

Shuy, Roger W. (1983). Three types of interference to an effective exchange of Infor-

mation in the medical interview. In The Social Organization of Doctor-Patient
Communication, S. Fisher and A.D. Todd (eds.). Washington: Center for Applied
Linguistics.
- (1984). Linguistics inother professions. AnnualReview of Anthropology 13:419-445.
Stubbs, Michael (1983). Discourse Analysis. Chicago, University of Chicago Press.
Tannen, Deborah and Wallat, Cynthia (1983). Doctor/mother/child communication:
Linguistic analysis of a pediatric interaction. In The Social Organization of Doctor-
Patient Communication, S. Fisher and A.D. Todd (eds.). Washington: Center for
Applied Linguistics.
Todd, Alexandra Dundas (1983). A diagnosis of doctor-patient discourse in the prescrip-
tion of contraception. In The Social Organization of Doctor-Patient Communication,
S. Fisher and A.D. Todd (eds.). Washington: Center for Applied Linguistics.
Treichler, Paula A.; Frankel, Richard M.; Kramarae, Cheris.; Zoppi, Kathleen; and
Beckman, Howard B. (1984). Problems and problems: Power relationships in a
medical encounter. In Language and Power, Cheris Kramarae, Muriel Schulz and
William M. O'Barr (eds.). Beverley Hüls: Sage.
Vargas, Donna Mayo (1984). Two types of legal discourse: transitivity in American
appellate opinions and casebooks. In Studies in Legal Discourse, B. Danet (ed.),
Texf 4(l-3):9-30.
West, Candace (1983). 'Ask Me No Questions. . .': An analysis of queries and replies in
physician-patient dialogues. In The Social Organization of Doctor-Patient Communi-
cation, S. Fisher and A.D. Todd (eds.). Washington: Center for Applied Linguistics.
Wodak-Engel, Ruth (1984). Determination of guilt: Discourse in the courtroom. In
Language and Power, Cheris Kramarae, Muriel Schulz and William M. O'Barr (eds.).
Beverley Hills: Sage.

Michael Agar is a professor in the Department of Anthropology, University of Maryland.

Although Most of his past work concerned patterns of urban U.S. drug use, he has
recently started a new study of independent truckers äs folk-theoreticians of small
business and regulation. Among his recent publications are The Professional Stranger: An
Informal Introduction to Ethnography and 'Toward and ethnographic language* in The
American Anthropologist.

Brought to you by | University of Arizona

Authenticated
Download Date | 6/2/15 9:17 AM