Journal of Sociolinguistics 4/3, 2000: 379±405

The interpreter as institutional

gatekeeper: The social-linguistic
role of interpreters in Spanish-
English
medical discourse1

Brad Davidson
San Francisco, California

Increases in immigration have led to an enormous growth in the number of

cross-linguistic medical encounters taking place throughout the United States.
In this article the role of hospital-based interpreters in cross-linguistic, internal
medicine `medical interviews' is examined. The interpreter's actions are
analyzed against the historical and institutional context within which she is
working, and also with an eye to the institutional goals that frame the patient
physician discourse. Interpreters are found not to be acting as `neutral'
machines of semantic conversion, but are rather shown to be active partici
pants in the process of diagnosis. Since this process hinges on the evaluation
of social and medical relevance of patient contributions to the discourse, the
interpreter can be seen as an additional institutional gatekeeper for the recent
immigrants for whom she is interpreting. Cross-linguistic medical interviews
may also be viewed as a form of cross-cultural interaction; in this light, the
larger political rami®cations of the interpreters' actions are explored.

KEYWORDS: Interpreting, medical discourse, medical anthropology,

discourse analysis, immigration, institutional ethnography

`Interpreters are the most powerful people in a medical conversation.'

Head of Interpreting Services at a major private U.S. hospital, May
1999.

1. INTRODUCTION
In this article, I examine the linguistic and social roles played by hospital-based
interpreters in medical discourse. The need for interpreters has become a fact
of contemporary medical practice; one study of 83 U.S. public and private
hospitals found that 11 percent of all patients required the services of an
interpreter (Ginsberg et al. 1995). Since Shuy (1976), a growing number of
researchers has become interested in close linguistic analyses of medical
discourse, but very little has been said to date about the linguistic and social

# Published by Blackwell Publishers Ltd. 2000

108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden MA 02148, USA.
380 DAVIDSON

role of the interpreter in cross-linguistic medical discourses. Yet these encoun

ters are common, and it is the interpreter, the only conversational participant
with the ability to follow both sides of the cross-linguistic discourse, who is
uniquely positioned within these discourses, to control the ¯ow of information
necessary for the achievement of the participants' medical and social goals.
In addition, in an era of massive population movements, the increase in
number and frequency of cross-linguistic medical encounters can also be
viewed as an increase in an institutional form of cross-cultural encounter
(Blackhall et al. 1995; Erzinger 1991; Marcos and Trujillo 1984; Martinez,
Lenoe and Sternback de Medina 1985; Phillips et al. 1996; Thamkins 1995),
with the interpreter acting as the point of negotiation and exchange between
the social contexts inhabited by the physician and the patient (Davidson 1998,
1999; Kaufert and Koolage 1984). In this article, then, I explore the contextually
and historically situated nature and role of the interpreter within these socio
medical interactions.
That there is a social component to hospital interpreting is itself a reasonably
uncontroversial, but largely unstudied, hypothesis. Hospital administrators and
physicians alike insist that it is both possible for, and the duty of, medical
interpreters to interpret without adding or subtracting meaningfully from the
content and intentions, and thus the eects, of utterances (to the degree to
which she does do so, she is considered incompetent). The problem lies in that
most, if not all, serious analyses of interpretation acknowledge that perfection in
interpretation is unattainable. At the very least, dierences in linguistic form lead,
inevitably, to dierences in meaning and reception, however small, and
semantically `identical' utterances in dierent languages vary greatly in their
social and contextual evaluation by speakers (e.g. Bendix 1988; Cartellieri
1983). In addition, the time constraints placed on interpreters forces them, in
the cases I observed, to do more than simply change the nuances of
utterances: they edit, and in some cases delete wholesale, conversational
oerings on a regular basis. There is considerable slippage, then, between how
the tasks that hospital personnel set for interpreters are believed to work in
practice, and the actual functions and linguistic actions that interpreters
perform.
The question remains as to what are the patterned ways in which the
interpreter in¯uences the discourses she interprets through these small, and in
some cases not-so-small, changes in linguistic form; what is the `interpretive
habit' of the socially positioned agents known as `interpreters' in a typical
medical encounter, and how do they conceive of their role in achieving
conversational goals? Interpreters interpret for a reason, because there is
some communicative or social goal that needs to be met; they do not simply
wander upon two speakers shouting at each other in dierent languages and oer
their services. From this point of view, the measure of the interpreter's success
may not be an abstract count of how `accurate' they are, but rather the degree
to which she allows, through her actions, the speakers ®rst to negotiate and
then to achieve their goals for the speech event in question.
# Blackwell Publishers Ltd. 2000
INTERPRETER AS INSTITUTIONAL GATEKEEPER 381

The mediated negotiation of conversational goals, however, is no trivial

matter; such goals are determined, for each conversational participant, by
historical contexts that frequently preclude any analysis of social equality
between the primary speakers. Many researchers have already noted that even
same-language medical discourse can be viewed as a form of interaction
between unequals, in which patients, as clients of the institution of the hospital
clinic, ®nd it dicult to establish a voice outside of the expected parameters of
medical practice (Mishler 1984; Waitzkin 1991; West 1984; Wodak 1996). In
the data examined in this article, the addition of a conversational mediator (the
interpreter) increases tremendously the patient's diculty in making herself, or
her agenda for the discourse, heard. The fact that the patients for whom these
interpreters are speaking are recent immigrants, mostly from the Third World,
highlights the fact that what interpreters are mediating in hospital discourse is
not only the diagnosis and care of patients, but also a form of cross-cultural
encounter between immigrants and agents of the institutions of the First World;
it is these agents who both provide services to these immigrants while
simultaneously educating them as to their role within the modern nation state
(cf. Gupta and Ferguson 1997).

2. BACKGROUND
Historically, most analyses of interpretation (the conversion of utterances from
one language into another) have been based on an oral model of translation
(the conversion of written texts), which has meant that most analyses of
interpretation have focused on monologues. Students of discourse have rarely
focused their attention squarely on interpretation itself, instead producing in
passing a tacit model of discourse interpretation as a sequence of discrete
linguistic conversions of isolated utterances. Interpreters are seen as conduits,
not conversational participants. For example, Hymes (1972), in his famous
SPEAKING mnemonic, lists the interpreter's role as that of `spokesman' or
`sender', but not as a `source' or `addresser'; similarly Goman (1981) calls the
interpreter an `animator', or one who makes noises, but not an `author' or
`principal', one to whom the meaning of utterances can be attributed (see also
Clark 1992, 1996).
But in interpreting a monologue one does not need to worry about turn
sequence, the rights of hearers to become speakers, or even the level at which
the primary speaker is being understood by the audience; monologues involve
one-way transmission, and the audience is largely unable to respond or to act
in setting the agenda for what will be discussed. However, the act of oral
interpretation of discourse is very dierent (cf. Roy 1999); it must take into
account all of these factors, but it is often reduced, in passing analyses, as the
interpreter's obligation to be a perfect echo of the primary interlocutors.

# Blackwell Publishers Ltd. 2000

382 DAVIDSON

2.1 The interpreter as conversational participant

Recently, however, students of language and discourse have turned their
attention to the nature and, to some degree, consequences of the interpreter's
role as an historical agent. This has led to a number of works that examine the
role of the interpreter or translator as linguistic and social intermediary (Bassnet
and Trivedi 1999; von Flotow 1997; Hatim 1997; Hatim and Mason 1989;
Rafael 1993; Snell-Hornsby 1988; Roy 1999; Venuti 1998; WadensjoÈ 1998).
All of these works share the common analysis that interpreters or translators,
far from `merely' converting and conveying the words of others, are centrally
employed in the work of mediating the achievement of conversational or
interactional goals, and that to a large degree responsibility for the achievement
of these goals lies squarely with the interpreter herself. Interpreters do not
merely convey messages; they shape, and, in some very real sense, create
those messages in the name of those for whom they speak. The context of the
interpreted speech event itself has also received considerable attention, and
the in¯uences of the social and historical facts surrounding an interpreted
speech event are seen to in¯uence greatly the interpreter's choices and the
resulting outcomes of the interaction (cf. Rafael 1993).2

2.2 Institutions and the mediation of post-colonial discourses

One signi®cant factor in¯uencing the manner and eects of interpretation is the
location of interpreted speech events within an historical-political timeline. With
the exception of business or diplomatic interactions, the majority of interpreted
discourses in the U.S. take place within the context of state sponsored or -run
institutions (the hospitals, schools, and judicial/legal system), between agents
of those First World institutions and the Third World immigrants who require or
are subjugated to the services provided. Institutional discourse is de®ned, in
large part, by the fact that institutionally de®ned goals and the institutionally
reinforced habits for achieving them provide clear signposts for how
communication should, and does, proceed, at least to those speakers familiar
with the institution in question (Bourdieu 1977; Cicourel 1983; van Dijk 1993;
Gupta and Ferguson 1997; Scheglo 1992; Wodak 1996; inter alia).
These institutional interactions thus stretch the notion of `neutrality' in
interpretation to the limit, as interpreters `are always placed in this contested
arena between being providers of a service and being agents of authority and
control.' (Candlin, in the introduction to WadensjoÈ (1998): xvii; italics in the
original). WadensjoÈ (1998: 68±69) writes:

`As do all professionalized intermediaries, interpreters work at providing a particular

service. Simultaneously, they ± of necessity ± exercise a certain control. Obviously
there is a potential con¯ict between the service and the control aspects, which
sometimes

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 383

surfaces in dilemmas reported in the literature on institutional communication. It

largely remains to be investigated how this con¯ict is handled in institutional
interpreter-mediated talk, where the gatekeeping is, in eect, doubled.' (italics in the
original)

Institutional interpretation has, then, a potentially disruptive social component

that cannot be ignored.
There have been several excellent studies of interpreted courtroom
discourse, all of which have pointed to the legal habit of selecting, privileging
and codifying certain utterances as `facts' (Scheppele 1989) as being the
dominant factor in determining how interpreters are allowed, and not allowed,
to interpret (Berk-Seligson 1990; Edwards 1995; Hewitt 1995; Mikkelson 1998).
In this article, I examine the `interpretive habits' of hospital interpreters through
a close analysis of their actions within the structured speech event known to
physicians as the medical interview; these habits are in¯uenced directly by both
the medical habit of dierential diagnosis and the institutional reality of chronic
time shortages in contemporary clinical practice.

2.3 Medical discourse and medical interpretation

There have been numerous, detailed studies of physician-patient discourse,
focusing primarily on the diculties patients and physicians have in commun
icating eectively with each other (Ainsworth Vaughn 1994; Cicourel 1983;
Frankel 1990; Hein and Wodak 1987; Mishler 1984; Robins and Wolf 1988;
Sarangi and Stembrouk 1996; Waitzkin 1983, 1991; West 1984; West and
Frankel 1991; Wodak 1996; inter alia). Most of these analyses center on one
type of medical discourse, the named speech event `medical interview'; this is a
structured, practiced interaction between the physician and the patient, taught
in medical schools, designed to quickly elicit the patient's complaint(s) so that
they may be diagnosed and treated. Medical interviews are thus a type of
verbal and physical investigation, a matching of unorganized experiences
against familiar patterns and processes of human vulnerability to disease. The
overt, elaborated goals of the medical interview are: 1) from the data provided,
determine what, if anything, is wrong with the patient; 2) elaborate a plan of
treatment for that ailment; and 3) convince the patient of the validity of the
diagnosis so that treatment will be followed. However, the elicitation of medical
`facts', or from another point of view the creation of medical facts through
medical practice, is heavily in¯uenced by a social evaluation of the meaning
and importance of whatever facts are thus uncovered or created (Foucault
1963; Waitzkin 1991); the practice of medicine, like the practice of interpreting,
has a social dimension that cannot be ignored.
Diagnosis is, then, an interpretive process in which the patient's physical and
verbal data is passed, by physicians, through a grid of medical meanings
(biological and social) and re-analyzed, so that `irrelevant' input from the
# Blackwell Publishers Ltd. 2000
384 DAVIDSON

patient may be excluded and the story of the disease constructed (Foucault
1963; Kleinman 1988; Mishler 1984). Indeed, this is the unmarked use of the
word `interpreting' in medical contexts; when I ®rst told the physicians at the
General Medical Clinic I would be conducting a study of interpreting, they
universally assumed that I meant a study of the ways in which physical and
verbal input are re-read as signs and symptoms of disease processes.
In fact, very little has been said about the concrete forms or eects of
interpretation of medical discourse. There are exceptions to this trend (Bendix
1988; Erzinger 1991; Marcos and Trujillo 1984; Martinez, Lenoe and Stern back
de Medina 1985; Weaver 1982), but the majority of literature written about
medical interpretation has come from two camps: physicians who use
interpreters (Baker et al. 1996; Baker, Hayes and Fortier 1998; David and Rhee
1998; Ebden et al. 1988; Putsch 1985; Vasquez and Javier 1991; Woloshin et
al. 1995), and interpreters themselves (Haner 1992; Juhel 1982; Kaufert and
Koolage 1984). The physicians generally lament the diculties of diagnosing
patients, establishing a clinical relationship, or providing adequate care to
patients when using an interpreter; the interpreters tend to focus on their role as
`linguistic ambassadors' for the patient, a stance in favor of overt `advocacy'
interpretation. Neither group, however, rests their arguments on analyses that
explore exactly how, in discourse, interpreters advocate or obfuscate the
conversational process.
It is not surprising that physicians have taken a recent interest in interpreta
tion: it would be hard to imagine a physician in practice or training today who
has not had to use an interpreter at least once to converse with a patient. At
Riverview General Hospital, the large, public county hospital in Northern
California where I conducted my research, the recent increase in interpretation
is readily apparent in Table 1. While Riverview may be unusual in the degree to
which multilingualism pervades everyday life, it is typical in the way that the
number of patients who need interpreters has increased in the last two
decades. It is also typical in that Spanish is both the most prevalent non-
English language, and in that it will remain so for the foreseeable future (cf.
Berk Seligson 1990). For this reason, in addition to the fact of my own bilingual
abilities in English and Spanish, Spanish language visits were chosen as the
subject of study.
Table 1: Riverview General Hospital patient demographics, by year

# of patients requesting % of total

Year # of patients seen interpreter patient population

1981 67,000 14,000 21 1993 133,000 53,000 40

Spanish-language visits in 1993: 34,000 (25% of all visits)

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 385

2.4 Methods and data

In the Spring and Summer of 1996, I conducted ®eldwork at Riverview General
Hospital's General Medicine Clinic (GMC), the outpatient unit of the internal
medicine division of the hospital. The patients there were undergoing treatment
for a variety of long-term illnesses, ranging from chronic back pain to stroke
rehabilitation to diabetes to congestive heart failure; their physicians were also
the primary care physicians within the hospital hierarchy, meaning that patient
referrals to other specialist physicians and clinics were orchestrated by these
internists.
Data collection centered on the ways in which hospital-based interpreters
were utilized within the clinic, how their presence during medical interviews
helped to shape the course and content of those interviews, and how they
mediated the potential clash of goals between the achievement of the overt
institutional goals of diagnosis and treatment that are set by standard medical
practice, and the not-necessarily identical goals held by the patients. I
approached the study as a political, social, and linguistic enterprise, with an eye
towards answering the following questions:
. What is the role of the interpreter within the goal-oriented, learned form of
interaction known as the `medical interview'?
. What is the `interpretive habit', and how does one engage in the practice of
interpreting?
. If interpreters are not neutral, do they challenge the authority of the
`physician-judge' (cf. Foucault 1979), and act as patient `ambassadors' or
`advocates' (as Haner 1992, Juhel 1982, and Kaufert and Koolage 1984
suggest); or do they reinforce the institutional authority of the physician and
the health-care establishment, and should we create a model for the
`interpreter-judge'? (cf. Foucault 1979)
In order to answer these questions, I observed both the interpreted medical
interview itself, and the institutional context thatsupported and gave meaning to
this speech event. Every physician and interpreter asked, and almost every
patient, agreed to participate in the study, most of them enthusiastically. It was
typical for patients, especially patients waiting for an interpreter to arrive, to tell
me that they thought it was an excellent idea for someone to study how
physicians talked to patients, largely because they thought it wasn't very well at
all.
The interpreters who form the focus of this article were professional in the
sense that they were paid employees of the hospital; none, however, had any
formal degree in interpretation or translation, and in this they appeared to be
quite typical of all of the hospital interpreters I have observed or spoken with in
Northern California. The speci®cs of training are dierent from hospital to
hospital in the San Francisco Bay Area, but in general they constitute nothing
more than a period of time following an interpreter on her daily rounds, an
# Blackwell Publishers Ltd. 2000
386 DAVIDSON

assurance that the interpreter in question is actually bilingual in the relevant

languages, and paperwork documenting that the interpreter is informed
(somewhat) about issues of patient con®dentiality. In this sense Riverview
General was `normal' ± interpreters were neither trained extensively nor
supported institutionally, and they performed their work in an ad-hoc vacuum of
accountability; the hospital monitored if they were present at a certain number
of physician-patient encounters, but expressed virtually no interest in
determining what they actually did in these encounters.
The data reported on here come from observations of over one hundred
patient visits, 50 of which were both observed and audiotaped. For those
interactions that were taped, all participants were requested to ®ll out a
questionnaire, and most were interviewed, at the end of the visit. The
audiotaped encounters included visits with both hospital-based sta and family
members acting as interpreters, and also monolingual interviews conducted all
in English or all in Spanish with no interpreter. There was a total of 10 Spanish-
English, professionally interpreted medical interviews taped. These 10 visits
were matched with 10 English-English visits, as closely as possible, for
similarity of patient, physician, and interpreter age, race, religion, and ethnicity,
and for nature of the interaction (®rst time visit, routine check-up) and the
patients' illnesses (diabetes, high blood pressure, etc.). The data presented
here, then, represent an exhaustive accounting of the data collected on
hospital-based interpreters speci®cally. The analyses of linguistic data below
are drawn from the set of 20 fully transcribed medical interviews, and from the
ethnographic and survey data collected on the clinic as a whole.

3. THE INTERPRETER IN MEDICAL INTERVIEWS

During the study at Riverview, one factor stood out as being overwhelmingly
contextually salient: the scarcity of time in modern medical institutions. The
amount of time patients spent waiting for their physician, the even longer
amounts of time spent waiting for interpreters, and the brevity of the physician
patient-interpreter encounter, added to the time constraints on modern medical
practice in general, seemed to be overriding factors in how interpreted medical
interactions took place. In myriad ways, all patients were shown that, from the
minute they entered the clinic until the minute they left, their time was not as
valuable as that of the physician, or of any other member of the clinic (cf. Elliot
1999). Patients who used interpreters had this message delivered in even
stronger terms. Often they were left alone in windowless examining rooms,
sometimes for up to an hour, while they waited for the interpreter to arrive. In
such cases the physicians would not wait, but rather would move on to the next
patient.

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 387

3.1 The interpreter as co-interviewer

One common scenario for interpreted medical interviews at Riverview was,
then, that the interpreter would arrive while the physician was busy elsewhere,
and she would begin some form of interaction with the patient before the
physician arrived. This had two eects. The ®rst was that, from the physician's
point of view, the process of elaborating a Chief Complaint (a named entity in
medical practice, usually written in chart notes as the abbreviated `CC') from
the patient was (apparently) simpli®ed; the interpreter might greet the physician
at the door of the examining room with an announcement of whatever the
patient had speci®ed as his or her problem, as in Excerpt 1, where 69 lines of
transcript have occurred before the physician enters the room. The second eect
was, however, that the interpreter thus set the focus of the initiation of the
interview, and would occasionally go so far as to conduct the initial portions of
the interview herself:

Excerpt 1 (from Visit 30):3

(Dr enters the room)
70 Pt Anda, a ver que dice el doctor.
Well, let's see what the doctor says.
71 Dr Hi:!
72 how are you doing?
73 Int Doctor, I was looking for something to put over there because he 74
wants to show you his:
75 (1.5 seconds)
76 foot but I didn't ®nd something.
77 Dr Oh.
78 Let's see:=
79 Int =One of those (xx)s, or.
80 Dr Maybe he: re, no:
81 Int Maybe in the (xx). ((banging noises ± searching for a stool)) 82 (2
seconds)
83 Dr Wouldn't surprise me.
84 (6 seconds)
85 Int At least we are not (xx).
86 Levante(te?) un poquito la pierna. (louder than previous English) Lift your
foot a little bit.
87 Pt SõÂ, sõÂ, senÄora.
Yes, yes, ma'am.
88 Ahora bien.
Okay now.
89 [x]
90 Int [¿CuaÂl] es el malo?
[Which] is the bad one?
91 Pt ¿Mande?
Excuse me?
92 Int ¿CuaÂl es el [enfermo?]
Which is [the `sick' (bad) one?]

# Blackwell Publishers Ltd. 2000

388 DAVIDSON

Â
93 Pt [E ste.]
[This one.]
94 Int A ver,
Let's see,
95 quitense el caletõÂn y el [XXX] por favor.
take o your sock and [XXX] please.
96 [(loud banging noise)]
97 Pt Oh, no.
98 (2 seconds)
99 Int He says that, ah,
100 you explained to him the last time then ah, . . .

It is impossible, in this stretch of text, to construe what the interpreter is doing

as `merely' conveying information. She is essentially running the interview, not
interpreting sequences of utterances. She asks the questions, and begins the
physical exam; her only interaction with the doctor is a request to help her ®nd
a stool for the patient to elevate his foot, and at line 99 the beginning of a
recapitulation to the physician of what was said in his absence. In taking charge
of the interview, she is preventing a potential initial greeting phase between the
physician and patient, and nowhere does she ask the patient, nor allow the
physician to ask the patient, what exactly has brought him to the clinic today.
Note that in line 70, the patient expresses an interest in hearing what his
physician has to say about his problem, but it is the interpreter with whom he
converses.
The interpreter has not, however, misunderstood the patient's earlier expres
sions of concern over his foot, which is, as becomes clear throughout the
interview, his true Chief Complaint; nor does the physician seem concerned
that the interpreter is conducting the preliminary physical exam. The only
problem, then, is that the interpreter has sacri®ced completely the notion that
the physician and the patient are participating, at this moment, in a
conversation with each other.
The patient and the physician appear to understand that the interpreter is not
interpreting, in the strict sense, but rather maintaining parallel and related
conversations that inform them, approximately, of the other's general verbal
oerings. They frequently make it clear when they want the interpreter to actually
interpret by telling her explicitly to do so. Excerpt 1 is taken up, below, a little
further along in the interview:

Excerpt 1 (from Visit 30), continued:

195 Int He says he feels:
196 good except his foot,
197 (.5 seconds)
198 ah: =
199 Dr =can I see his other foot?
200 Int A ver,
Let's see,

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 389

201 Pt DõÂgale que,

Tell him that,
202 Int Y, y it stops when, when,
203 it's worse, when he walks.
204 (3 seconds)
205 Pt Y yo, ya sõÂ me siento bien.
And I, now I feel good.
206 (2 seconds ± Dr takes o Pt's sock and looks at other foot). 207
Ah, (que esos?) no los pongo, si no los (pide?).
Ah, (that these?) I won't wear these, if he doesn't (ask?).
208 Int Oh my god, it's totally (xx) ± (very softly, aside to Dr). 209
(1.5 seconds)
210 Pt DõÂgale que eso no me duele, eso es solo como una reventadita.
Tell him that this doesn't hurt me, it's like a little eruption.
211 Int He says that one is nothing, and it's like a little,
212 (xx?)
213 (3 seconds)
214 ¿Pero se acuerda que asõ le empezo el otro?
But do you remember that the other one started this way?
215 Pt SõÂ, asõ como esta eÂse.
Yes, just like this one is.
216 Que bien que Ud. se acuerde.
How good that you remember.
217 Int I mentioned that remember then the other one start the same way. 218 Dr
I know. ((e.g. that the other foot started the same way)) 219 Int And he said, yeah,
that's good that you remembered that.

The interpreter is still not following a model of strict sequential interpretation:

the only straightforward sequence of utterance-interpretation happens in lines
210±212. Even a request for interpretation may not be granted. In lines 201
and 210, the patient expressly asks the interpreter to interpret his words by
prefacing his statements with dõÂgale, `tell him', but the patient's subsequent
oering after line 201, presumably the `I feel good now' in line 205, is not
interpreted. It is not only the patient who has his oerings left untranslated,
however: line 199 from the physician is not put into words, either, although his
request to `see [the patient's] other foot' is ful®lled. Notice also that the
interpreter, who knows the patient from past interactions, comments upon the
state of the patient's foot as being the same way that the other foot started to
have problems (line 214), a comment which elicits a response from the patient
directed at her (`Ud.', line 216) and not the physician; only when this
interaction, initiated by the interpreter herself, is completed does she recount
what she has said to the physician.
The patient's problem is not minor: he is in danger of having his toes
amputated as a result of complications from unmanaged diabetes. The
physician knows this, the interpreter knows this; the one conversational
participant who does not seem to grasp the severity of the problem is the
# Blackwell Publishers Ltd. 2000
390 DAVIDSON

patient himself, who at the end of the interview remains unconvinced by the
physician's warnings that if his diet does not change he will lose, not only his
toes, but eventually his feet as well. The patient manages to get a referral to
the wound management clinic to have his feet cleaned, which was his initial
goal; but if one of the institutional goals of a medical interview is to not only
elaborate a diagnosis and plan of treatment, but also to convince the patient of
the validity of the diagnosis and plan, this interview has failed in that the patient
leaves completely unconvinced that his problems are as severe as the
physician and interpreter tell him they are. It is not possible to say with certainty
that this is the result of the interpreter's actions; what is clear, however, is that
the patient's voice is signi®cantly modi®ed by having to speak through an inter
preter, and that the interpreter is frequently speaking, not as an echo, but in her
own voice.

3.2 Quanti®able patterns of interference in interpreted medical interviews

One could ask at this point, as did many of the physicians who took part in this
study (acting perhaps as devil's advocates), what is the harm in the interpreter
assisting the physician in conducting the medical interview, especially if time is
short and it speeds the interview process along? The ®rst response would be
that nowhere is it stated that speed is the primary goal of a medical interview;
these are institutional constraints, but they are universally decried as having a
detrimental eect on the physician-patient experience. Time is scarce in
hospitals today, however, and interpreters are conscious of their role as
facilitator and editor; during one interaction (visit 11), after several minutes of
conversation with the patient in the absence of the physician, the interpreter
looked at me and said, `you chose one that's hard to keep on track'. The
patient had been providing a detailed history of the diculty he had had at
various clinics aliated with Riverview General. The interpreter's statement made
it clear that she felt this was extraneous information, and that it was her job to
keep the patient on track, as measured against what she believed to be
relevant information for a medical interview. It was not clear that the patient's
narrative was, however, irrelevant; the `Social History' relating to an illness is
part of the routine medical interview, coded in patient charts under the heading
`SH'.
The consistent attempt to keep patients `on track' led to a number of
quanti®able phenomena in the discourse. Tables 2 and 3 show how patient
generated direct questions were dealt with in the two sets of interviews. For
both sets of interviews, almost all of the direct questions asked were answered.
However, for the patients using an interpreter over half (18/33) of all of the
questions which were directed at the physician were answered by the inter
preter, without the physician ever hearing the question. The signi®cance of this
pattern of short-circuiting question-and-answer sequences between patients
and physicians is not only that patients are receiving answers from their
interpreter and not their physician; it is also that physicians have no idea
# Blackwell Publishers Ltd. 2000
INTERPRETER AS INSTITUTIONAL GATEKEEPER 391

Table 2: Treatment of patient-generated direct questions in 10 same-language

visits

# of questions Answered by Not answered by from patient physician

physician

Total 55 53 2

Table 3: Treatment of patient-generated direct questions in 10 interpreted

visits

# of Not
questions Passed on passed on Answered Answered Not from pt to Dr to Dr
by Dr by interp answered

Total 33 15 18 12 17 4

that their interpreted patients are asking questions at all, which increases the
likelihood that these Hispanic patients will be seen as `passive' (cf. Baker,
Hayes and Fortier 1998; Erzinger 1991) and also prevents the physician from
following up on dicult questions or questions that display a deep misunder
standing, on the part of the patient, as to what the diagnosis or plan of
treatment are.
Another possible analysis for this treatment of direct questions is that these
questions pose a threat to the physician's authority within the medical interview
(Ainsworth-Vaughn 1994). In medical interviews it is the physician, and not the
patient, who typically asks the questions (cf. Mishler 1984). We have already
seen in Excerpt 1 that interpreters are themselves capable of producing spon
taneous requests to patients, thus taking on the physician's right to ask
questions; the interpreters' habit of answering questions might be viewed as a
move to insulate the physician, and thus the institution of the clinic, from patient
challenges to its authority. We will return to this issue in Section 3.3, below.
In a related vein, patients' physical complaints themselves, the raison d'eÃtre
of the medical interview, are often lost in the conversational shue. Table 4
shows the number and content of identi®able patient complaints from two
matched interviews, interviews 6 and 7. They have been chosen for detailed
comparison because of the large number of similarities they share: both
interviews took nearly the same time, and dealt with `typical' patient complaints
at the GMC ± chronic discomforts that did not appear life-threatening and which
were dicult to diagnose and subsequently treat. Both interviews took place on
the same day, with the same physician, and the patients were roughly the same
age. Both the English speaking patient in visit 7 and the Spanish speaking
patient in visit 6 produced new complaints, in addition to a number of vaguely-
de®ned and
# Blackwell Publishers Ltd. 2000
392 DAVIDSON

dicult to treat conditions that had been addressed in previous visits. The two
visits were also representative in that, given roughly similar clinical scenarios,
in comparison to the same-language visit the interpreted visit was marked, as
were nearly all of the observed interpreted visits, by severe communicative
diculties. These were of a type and to a degree that far surpassed the `normal'
communicative diculties encountered by same-language patients and reported
in research on physician-patient discourse.
The complaints oered by the English speaking patient were all addressed
directly, in one form or another, and nearly all of them were diagnosed and
treated; for the Spanish speaking patient, however, most of his complaints were
left undiagnosed and untreated, most signi®cantly the one complaint that he is
most concerned with (see Excerpt 2, Section 3.3). This was due to one of three
processes: to the physician not hearing the complaint because the interpreter
didn't pass it along; to the physician hearing the complaint but not addressing it,
leaving the interpreter with nothing to say to the patient; or to the physician
hearing and addressing the complaint, but the interpreter not passing the
physician's commentary on to the patient. The majority of the patient's
complaints, in visit 6, were left without a concrete or even partial diagnosis or
plan of treatment. Notice also that the ®nal diagnosis for most of the Spanish
speaking patient's complaints are `general pains, which the physician sees as
related to his ``mood'' '; his illnesses are considered psychosomatic, to a large
degree, which is common for patients who speak Spanish (Erzinger 1991;
Marcos and Trujillo 1984). The English speaking patient, a recovering
intravenous drug abuser, was no less depressed, but his complaints of physical
discomfort were taken seriously enough by the physician to have them
addressed individually and concretely. One of the most negative eects of

Table 4: Complaints addressed and diagnosed in visits 6 and 7

Diagnosis supplied Treatment suggested

Complaints: visit 6, Spanish

Vision (new) No No Foot pain No No Arm/hand pains Yes No General pains
Yes Yes Mood No Yes
Complaints: visit 7, English
Wrist pain (new) No Yes Cough (allergy) Yes Yes High blood pressure Yes Yes
Frequent urination (new) Yes (possible diagnosis) Yes Prostate node (new) Yes Yes
(urologic referral)

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 393

interpretation at Riverview, in fact, is the tendency for physicians to see

patients who can't speak for themselves, as a result of the conversational
diculties, as `cranks' or patients who complain of phantom problems (Davidson
1998). What is left to examine is the role played by the interpreter in interview
6, to determine what part she played in thwarting the elaborated goals of
diagnosis and care delivery.

3.3 The loss of patient complaints

In Excerpt 2 we see portions of the interview in visit 6, which took place
between a Spanish speaking male patient in his mid-40s, an English speaking,
male, Anglo physician in his mid 30s, and a Spanish-dominant, professional
female interpreter in her mid 40s (a dierent interpreter from that in Excerpt 1).
This is not the ®rst visit between the patient and the physician, who have
known each other for three years; the interpreter, too, knew the patient, as this
was the `third or fourth' time she had interpreted for him. The excerpt begins
after the patient and the physician had already had a chance to interact, brie¯y,
an interaction that I observed; the patient had enough English to say his `eyes
hurt', but he could say no more than this, at which point the physician decided
to call the interpreter (why, after a 3 year clinical relationship, the physician
decided to call the interpreter after trying to converse with the patient, remains
unclear). The patient had been left waiting in the examining room for over 45
minutes, as the physician moved on to his next patient rather than wait for the
interpreter to show up.
The excerpt begins, then, with the physician and the interpreter (and myself)
arriving in the exam room to see the patient, who has at this point been waiting
alone in a windowless room. Notice that the interpreter is licensed, in the
physician's ®rst turn, not simply to interpret, but rather to explore `what did [the
patient] mean by this?' (line 23) when he said his `eyes hurt'. From the very
beginning of the interaction, she becomes responsible, then, for not only
conveying information, but for ®rst collecting it, in immediately usable fashion,
in the name of the physician. The interpreter's subsequent actions should be
judged in the light of this request from her institutional superior, the physician
for whom she is interpreting, to actively clarify the patient's verbal output.
Notice also that, from the beginning, we hear the physician, patient, and
interpreter struggling to construct a coherent account of the patient's Chief (or
at least initial) Complaint about his eyes, but that they fail to establish anything
more than that the patient's eyes have a problem relating to `burning', `tearing',
and cloudiness or a possible complete loss of vision. The interaction was
perhaps even more muddled than the transcript shows; when the interpreter
struggles to convey exactly what the patient has said, it is because what he is
saying is not entirely clear, a fact which is not commented upon by either the
physician or the interpreter. The problem of de®nition, central to medical
diagnosis, is never resolved:
# Blackwell Publishers Ltd. 2000
394 DAVIDSON

Excerpt 2 (from visit 6):

15 Dr Mr. X was telling me ± no, come sit here.
16 We have all these chairs, no there's two chairs.
17 Grab a seat.
18 Ok,
19 Int mm-hm.
20 Dr So: he was telling me tha: t
21 he was having problems no: w, he said, with his vision. 22 He
said sometimes he can't see at all.
23 What did he mean by this?
24 Int Mm-hmm.
25 Dice el doctor que esta teniendo problemas con la vista The doctor
says that you are having problems with your vision
26 que unas veces
that some times
27 no puede VER?
you can't SEE?
28 Pt Ah, sõÂ, sõÂ, xx en la: la vista, se me,
Ah, yes, yes, xx in the:, the vision,
29 empieza a salir agua
water starts to come out
30 aunque estoy en xxx
even though I am in xxx
31 y-y: como que tengo chile allõÂ
and-and like I have chili there
32 Int Eyes get teary,
33 and, burning, feels like,
34 hot chili.
35 Dr Hot chili. But it's not that his eyes go black
36 it's that his eyes ar: e.
37 Int Pero no es que la vista se le
But it is not that your vision
38 se le ponga totalmente oscura, negra.
goes completely `obscure' (opaque), black.
39 Pt Bueno: =
Well: =
40 Int =Es simplemente que le arden los ojos.
=It's simply that your eyes burn.
41 Pt A veces se me pone, se me va la vista.
Sometimes it gets, sometimes my vision goes.
42 Cuando pasa esto se me va la vista.
When this happens my vision goes.
43 Int Se le va la vista=
Your vision goes=
44 Pt =SõÂ
=Yes
45 Int ah:, ¿que muy oscuro?
ah:, what very dark?

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 395

46 Pt Muy oscuro.
Very dark.

One aspect of this transcript fragment seems immediately apparent, which is

that this opening phase of the interview, which phase typically concerns the
elicitation and elaboration of the patient's complaint, neither clari®es nor
furthers the physician's understanding of what, exactly, is wrong with the
patient's eyes. To this point in the transcript, very little has been established
beyond the fact that the patient's eyes hurt him, they may go dark, and they
burn. The doctor has entered, in his attempt to clarify the patient's complaint,
what the colonial missionary Father Murillo, cited in Rafael, referred to as ```a
labyrinth without a clue'' . . . beset with digressions and non-sequiturs' (Rafael
1993: 133); while Rafael is referring to the diculty in hearing cross-linguistic
confessions from native Tagalogs in the colonial Philippines, the phrase could
easily apply to the physician's attempts at clari®cation and de®nition of the
patient's problem with his eyes.
 The confusion that is evident in the transcript was equally evident in the
actual interview; the physician was visibly upset that he could not get a clear
picture from the patient of what was wrong, and the patient was also visibly
upset that he was asked the `same' question over and over. Clark (1992, 1996)
describes the `achievement' of a contribution to a discourse as the moment
when both parties understand what has been said, and believe that the other
also understands; at this point in the excerpt, nothing substantial that would aid
in diagnosis has been achieved in the medical interview.
It should be noted here also that the interpreter's changes to the dialogue are
not entirely related to problems inherent in the act of linguistic conversion itself.
Her insertion in line 40, for example, of the evaluative adverb simplemente,
`simply', to modify the question about the patient's eyes burning, is a judgment
that is hers and hers alone. The physician, by his negative question, implies
this relative scale of severity (burning eyes are less serious than loss of vision),
but it is the interpreter who puts this implication into concrete form. In addition,
her follow-up question to the patient at the end of the fragment, in line 45,
represents a small but signi®cant departure from the notion that the interpreter
conveys all and only what was said; she is asking the patient to clarify, to her,
what he is saying, before she attempts to pass this information along to the
physician. In the end, lines 47±83 do not serve to pin down the complaint, and
the physician ®nally moves on. In lines 84±85, below, he turns from the
problem of de®nition of symptoms to the question of duration of these (still
vaguely de®ned) symptoms, asking, `Ok, so, how often does this happen?'
To this point in the encounter the patient can infer from the interpreter's
oerings that both he and the physician understand that he is concerned about
his eyes, and that the physician now shares this concern. But when the patient
reports that his eyes burn and get teary, the question that is addressed to him
in
# Blackwell Publishers Ltd. 2000
396 DAVIDSON

reply to this report is the negative, `but it is not the case that you lose your
vision' (lines 37±38). What may be inferred, by the patient, from this question is
that the physician has taken the patient's reported symptoms, not as a positive
armation of illness, but rather as a negative armation of a more serious illness.
The patient then states that he does lose vision when his eyes tear and water,
although he then modi®es that, after being led by the interpreter (line 45), to
agree that, instead of losing sight, his vision goes `very dark'.
I do not wish, here, to overstate the analyses of what the conversational
participants are attempting to achieve with their turns-at-talk; it is not possible to
say, with absolute certainty, the goals of the dierent conversational oerings that
each participant proers. Epistemic and emotional states are not available for
de®nite analyses, and it is impossible to fully catalogue the intended eect of
utterances. It is possible to claim, with some very high degree of certainty, that
the interview in Excerpt 2 has very quickly become bogged down in an attempt,
on the part of the physician, to determine what exactly the nature of the Chief
Complaint is, centering, not on a dierential diagnosis, but on the establishment
of an agreed upon set of symptoms that the patient is reporting. With this in
mind, it is noteworthy that the physician chooses to move on in the interview at
all, from de®ning the complaint to ®nding out about the temporal markers of
diagnosis, that is, frequency, intensity, and duration of the (as yet unde®ned)
symptoms (lines 84±85, below).
The next transcript fragment from visit 6 shows how the question of the
patient's eyes is resolved within the interview. Here the interpreter's role as co
diagnostician comes to the fore: the physician asks for a further clari®cation of
time-of-onset of the complaint, to which the patient answers in an indirect, but
entirely relevant, fashion. The interpreter, however, ignores the patient's
oerings in lines 105±111, and instead re-tracks him to give a strictly temporal
answer to the physician's initial question:

Excerpt 2, continued:
84 Dr Ok, so,
85 And how often does this happen?
86 Int Uh, ¿cuaÂnto le sucede esto?
Uh, how often does this happen to you?
87 Pt Pues, uh:,
Well, uh:,
88 unas dos veces yo creo al mes,
about two times I think a month,
89 me sucede.
it happens to me.
90 Int two times
91 [a month.]
92 Pt [maÂs o menos]
[more or less]
93 Int [more or less.]

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 397

94 Dr [Twice a month.] For how long?

95 Int ¿Por cuaÂnto tiempo le [(xxx)]?
For how long does it [(xxx)] (to you)?
96 Pt [me dura como:]
[it lasts (me) like:]
97 (1 second)
98 a veces me dura casi media hora.
Sometimes it lasts almost a half hour.
99 Int Sometimes he takes, ah, i-it lasts
100 uh-h, half an HOUR when it happens
101 Dr Ok, right
102 So::
103 And how long has it been going on for?
104 Int ¿Y por cuaÂnto tiempo le ha venido sucederle esto?
And for how long has this been happening to you?
105 Pt Pues, yo trate de decirle al doctor de
Well, I tried to tell the doctor
106 de hace maÂs,
more than,
107 cuatro cinco visitas para atraÂs
four ®ve visits ago
108 Int mm- [hm]
109 Pt [que] ya me estaba sucediendo.
[that] it was already happening to me.
110 Pero:,
But,
111 Que no se si eÂl me entendõÂa o no.
I don't know if he understood me or not.
112 Int Pero, ¿hace,
But, since,
113 ha-hace cuaÂndo que le comenzo a,
Si-since when did this start,
114 a suceder esto?
to happen to you?
115 Pt Mas o menos como un anÄ o, yo creo.
 More or less about a year, I think.
116 Int About a year.
117 Dr Ok. (9 second pause)
118 And it goes away by itself?
119 Int Ah:
120 y, it.
121 AsõÂ como le viene esta molestia,
And as this discomfort comes to you,
122 se le quita sola.
it goes away by itself?
123 Pt Me quita, sõÂ.
It goes away (from me), yes.
124 Int Yes, it goes away by itself.

# Blackwell Publishers Ltd. 2000

398 DAVIDSON

Lines 101±116 are critical to this analysis. After a string of successful closed
questions (lines 85±100), the physician asks for how long the patient has been
suering these (uncertain) symptoms (lines 101±103), and the interpreter relays
this question (line 104). But the answer, from the patient, is not a direct one:
instead he replies that he has been trying to tell the physician for quite some
time that his eyes are bothering him, but that he is unsure if the physician has
heard (and by implication understood) him or not (lines 105± 111).
The reply is indirect, but relevant. The patient is not simply replying to the
question, but is rather addressing himself to the inferable basis of the question;
this is that a commonly-held and agreed upon medical fact has not been
established, despite the patient's repeated attempts to establish it over time.
The patient does not believe that the physician has assimilated his complaint
into the medical record, and the patient has every reason to suspect this is the
case given the nature of the physician's question in line 103. The patient does
not know if a crucial piece of information has been accepted by the physician,
and suspects, correctly, that it has not, because he has never had a reply, in
this or prior visits, that would make it possible for him to infer that this is the
case. The interpreter's subsequent action, which is an attempt to re-focus the
dialogue on the immediate semantic basis of the physician's original question
(lines 109± 114), is met with a reply of `one year' (line 115).
The interpreter's verbal actions in lines 104±116 are critical. Having been
asked to initially determine `what the patient meant' by his complaint about his
eyes, the interpreter has moved on to determining the relevance of an
utterance to the process of diagnosis at hand: the physician has asked a
question, presumably in anticipation of a strict temporal reply, and the patient
answers with a more complicated, albeit entirely relevant, answer about the
nature of the question itself. The interpreter here evaluates the patient's
response and dismisses it as irrelevant (`but . . .', line 112) to the initial closed
question, denying its entry into the discourse. The interpreter is acting as pre-
®lter for patients' utterances, screening them for relevance to the physician's
questions: as noted earlier, however, converting data by passing it through a
grid of medical meanings is the central component of the process of diagnosis
itself. In addition, it is entirely possible that the interpreter here is not merely
screening the patient's answer for relevance, but that she is deleting it
wholesale, to protect the physician and the institution of the hospital from the
critique that the patient's complaint has been repeatedly ignored.
This maneuver eectively obliterates the chance that in this visit, as in prior
visits, the patient will be able to establish the medical fact that, not only do his
eyes bother him, but that he has been attempting to report this problem for
quite some time. With respect to patient complaints in general, the physicians
who took part in this study often spoke about the importance of determining,
not only the exact complaint, but why the complaint has become signi®cant or
urgent enough to be brought up by the patient now. The fact that the report has
# Blackwell Publishers Ltd. 2000
INTERPRETER AS INSTITUTIONAL GATEKEEPER 399

been made repeatedly, over time, is important, because it supplies the

physician with the information that this complaint is neither trivial nor recent.
The physician, however, does not question the interpreter with regards to what
the patient has said in lines 106±111; he hears a coherent reply to a closed
question, and moves on.
Far from being a valid CC, the report of burning eyes now becomes a
somewhat trivial complaint, for how bad could it be if a year has passed and
only now the patient is bringing the symptom to the physician's attention? The
physician's subsequent question, in line 118, adds credence to the analysis
that, because the physician receives a strictly temporal reply via the interpreter,
he no longer takes the patient's complaint seriously: the question, `it goes away
by itself?', may be read, in a medical context, as the question, `is this ailment
self limiting (e.g. self-correcting), and do I need to seriously address it?' Self-
limiting ailments, such as colds (which go away in a few days regardless of the
medical care delivered or withheld) are generally considered non-issues by
medical practitioners, because there is nothing that can be done, medically, to
®x them. When the reply comes back that yes, the symptom cures itself (line
124), the physician quickly works, over the next 50 lines of transcript, to move
the interview past the report. When the physician hears, ®nally, that perhaps
the patient's need of glasses may in the future give permanent relief, he resets
the interview with an open question, still in search of a valid CC upon which to
focus:

Excerpt 2, continued:
179 Dr Oh, good.
180 (8 seconds).
181 And:,
182 Is there any, other,
183 main thing that is bothering him today?
184 I know there's a lot of problems,
185 but if there was only one
186 other thing he was gonna tell me about today, what would he choose?

Notice that the question in lines 182±186 can be read as serving several
functions: the ®rst, of course, is to re-set the interview, so that a `valid' Chief
Complaint may be identi®ed and addressed. Another, however, is that the
physician here is himself trying to keep the patient on track, by asking him to
pre-evaluate, on his own, what is most medically important. One ®nal message
which might be read into this question is that the physician may be serving
notice to the patient that he has wasted a certain amount of time with his
complaint about his eyes, and that there is now time for only one more of his
complaints to be addressed before the physician will close out the interview.
The problem is that the patient's problem with his eyes is, in fact, a valid Chief
Complaint. The interpreter's pre-evaluation and de-facto editing of the patient's
contributions to the interview result, in this case, not in keeping that patient `on
# Blackwell Publishers Ltd. 2000
400 DAVIDSON

track', but rather in un-tracking the achievement of the institutional goals

(diagnosis and treatment) of the interview itself.
4. CONCLUSIONS AND DISCUSSION
In this article I have outlined the role of interpreters in one form of medical
discourse, the internal medicine `medical interview'. The linguistic data, both
quantitative and qualitative, points strongly away from a conclusion that
interpreters are acting as `advocates' or `ambassadors' for interpreted patients,
but are rather acting, at least in part, as informational gatekeepers who keep
the interview `on track' and the physician on schedule. While the interpreters do
in fact convey much of what is said, they also interpret selectively, and appear
to do so in a patterned (non-random) fashion. There is no evidence in the data
presented here (nor in the larger data set) of interpreters putting forth the
patient's agenda vigorously, as is claimed by Haner (1992) and others. This is
not the inevitable role that interpreters must take in hospital discourses,
however, and the reasons they act in this way at Riverview is largely a result of
their position within the hospital hierarchy.
The practice of medical interpreting is not highly valued within the hospital
clinic; when I began my study, I was told by a sympathetic physician that he
had also been interested in studying interpreters, but had been told by a
hospital administrator not to do `any studies that tell me I need to hire more
interpreters; we can't aord the ones we have now'. There were only seven full-
time Spanish English interpreters at Riverview General Hospital, not nearly
enough to take care of the 33,000 patients who needed Spanish interpretation,
even given the large number of bilingual physicians and family-member
interpreters utilized in individual clinics. The training given to these interpreters
was scant; the requirements for becoming an interpreter at Riverview were a
good grasp of both English and Spanish, and the ability to translate 50 medical
terms on a test with complete accuracy. There was no training in discourse
processes, and the training for how medical interactions worked was on-the-
job. Physicians, for their part, received absolutely no training in how to use
interpreters, beyond being told how to call them to come interpret.
The clinic sta were also consistently wrong in their predictions of who would
need an interpreter, and more often than not would be forced to call an
interpreter for an unscheduled interpretation, rather than scheduling in
advance. Conse quently, the interpreters were always running behind,
postponing scheduled interpretations and answering pages through the day
that added to a large list of patients who needed their services. In addition,
during my study over 100 nurses were ®red at the hospital, at the same time
that the physicians themselves were being asked to see more and more
patients in a shorter and shorter time period.
These time pressures all gave rise to competing mandates for the
interpreters. Institutionally, they are ocially required to act as an `instrument',
saying all and only what has been said; in practice, however, they are
encouraged to keep
# Blackwell Publishers Ltd. 2000
INTERPRETER AS INSTITUTIONAL GATEKEEPER 401

the interview short, and to keep patients `on track'. In competition, it was almost
always the latter requirement that won out, and interpreters frequently engaged
in furthering the physician's perceived agenda for the discourse. This
happened, not only because of time pressures, but because hospital based
interpreters are, in the end, members of the hospital community where they
work and interact daily; they are institutional insiders, and ally themselves as
such.
The larger rami®cations of the interpreter's role in medical discourses are
also signi®cant. Of the power of institutional encounters to de®ne citizens'
relations with the state itself, Foucault writes (1979: 304):
 We are in the society of the teacher-judge, the doctor-judge, the educator-judge, the
``social-worker''-judge; it is on them that the universalreign of the normative is
based . . .

All that is needed to make this quote perfectly relevant to the analysis at hand
is to add the words `the society of the interpreter-judge', for it is interpreters
with whom and through whom recent immigrants interact with institutions of the
state.
Interpreters are not, and cannot be, `neutral' machines of linguistic con
version, both because they are faced with the reality that linguistic systems are
not `the same' in how they convey information contextually, and also because
they are themselves social agents and participants (albeit special ones) in the
discourse. It is possible for them to interpret evenly, however, and it is not the
case that professional, hospital based interpreters need to work as an extra
gatekeeping layer through which patients must pass in order to receive medical
care. One could argue, as I would, that the interpreters' wholesale alignment
with the institution of Riverview General Hospital (which is, not coincidentally,
their employer) is both unethical and a truly poor form of interpretive practice.
As stated earlier, however, it is the context of communication that is
fundamental in de®ning how the interpreter will carry out her role, and how she
should be judged in that role; given that the physicians' command was, ®rst
and foremost, to keep the interview short, interpreters at Riverview may in fact
be doing a good job at a bad task. The real issue is that they are doing a job
that is dierent, in daily practice, from the job they are typically assumed to be
doing. This means that they are not trained, nor licensed within the institution
(i.e. they cannot write referrals or prescriptions, and may not make notes in the
patient's permanent record), to do the things they are in fact doing (collection
and analysis of data; establishing a `therapeutic rapport' with the patient); nor
are they given any form of institutional support for the true nature of the work
that they do. The construction of the interpreter as a simple instrument of
semantic conveyance is only possible when those who hire and use
interpreters imagine that it is possible for interpretation to be the task of merely
echoing content faithfully. It is this conceptualization of the interpreter's work
that renders her daily practice of acting as co-diagnostician invisible, which in
turn engenders a vacuum of responsibility, both within the discourse and with
respect to the delivery of health care to non-English speaking patients in
# Blackwell Publishers Ltd. 2000
402 DAVIDSON

general. This slippage, between what interpreters are asked, ocially, to do, and
what they are really doing in daily practice, allows the practice to continue
unmonitored and unevaluated; the invisible nature of the interpreter's role as
co-diagnostician is the eect, rather than the interpreter's incompetence being
the cause, of the broad dissatisfaction physicians and patients at Riverview
express towards medical interpreting in general.

NOTES
1. The research for this article was conducted with the aid of a Doctoral Fellowship from
the Department of Linguistics at Stanford University; part of the preparation for this
manuscript took place with the aid of a Postdoctoral Fellowship at the Stanford
University Center for Biomedical Ethics, funded by the Robert Wood Johnson
Foundation's `Last Acts' Campaign. I would like to thank Stuart Tannock, Elizabeth
Traugott, John R. Rickford, Penelope Eckert, Elizabeth Armstrong Davidson, and the
members of the Sociolunch group at Stanford University for their insights. All faults
 are entirely my own.
2. In this view, one might call interpretation the pre-eminently contextual linguistic act, or
at least the most consciously contextual of linguistic behaviors. 3. Transcription
Conventions:
[ ] overlapping turns
= simultaneous beginning of one speaker's turn/end of another's turn (latch ing)
: lengthening
CAPS loud
Italics translation of Spanish
(text?) dicult to hear
(noise) description of non-verbal noise
(xx) impossible to hear; each x is one syllable, if syllables can be discerned
((text)) description of physical actions, or meta-commentary on discourse

REFERENCES
Ainsworth Vaughn, Nancy. 1994. Is that a rhetorical question? Ambiguity and power in
medical discourse. Journal of Linguistic Anthropology 4: 194±214.
Baker, David W., Ruth M. Parker, Mark V. Williams, Wendy C. Coates and Kathryn
Pitkin. 1996. Use and eectiveness of interpreters in an emergency room department.
Journal of the American Medical Association 275: 783±788.
Baker, David W., Risa Hayes and Julia Puebla Fortier. 1998. Interpreter use and
satisfaction with interpersonal aspects of care for Spanish-speaking patients. Medical
Care 36: 1461±1470.
Bassnett, Susan and Harish Trivedi. 1999. Post-colonial Translation: Theory and
Practice. London and New York: Routledge.
Bendix, Edward H. 1988. Metaphorical and literal interpretation: Cross-cultural com
munication in medical settings. CUNYForum 13: 1±16.

# Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 403

Berk-Seligson, Susan. 1990. The Bilingual Courtroom: Court Interpreters in the Judicial
Process. Chicago and London: The University of Chicago Press.
Blackhall, Leslie J., Sheila T. Murphy, Geyla Frank, Vicki Michel and Stanley Azen. 1995.
Ethnicity and attitudes toward patient autonomy. Journal of the American Medical
Association 274: 825.
Bourdieu, Pierre. 1977. The economics of linguistic exchanges. Social Science and
Information 16: 645±668.
Cartellieri, Claus. 1983. The inescapable dilemma: Quality and/or quantity in interpret
ing. Babel 29: 209±213.
Cicourel, Aaron V. 1983. Language and the structure of belief in medical commun
ication. In Sue Fisher and Alexandra Todd (eds.) The Social Organization of Doctor
Patient Communication. Washington, D.C.: Center for Applied Linguistics. 221±240.
Clark, Herbert H. 1992. Arenas of Language Use. Chicago: University of Chicago Press/
Center for the Study of Language and Information.
Clark, Herbert H. 1996. Using Language. Cambridge: Cambridge University Press.
David, Rand A. and Michelle Rhee. 1998. The impact of language as a barrier to eective
health care in an underserved urban hispanic community. The Mount Sinai Journal of
Medicine 65: 393±397.
Davidson, Brad. 1998. Interpreting medical discourse: A study of cross-linguistic
communication in the hospital clinic. PhD Dissertation. Department of Linguistics,
Stanford University.
Davidson, Brad. 1999. Dialogue in cross-linguistic medical interviews: The interpretation
of interpretive discourse. Proceedings from the Sixth Annual Symposium on
Language and Society, Austin.
Diaz-Duque, Ozzie. 1989. Communication barriers in medical settings: Hispanics in the
United States. International Journal of the Sociology of Language 79: 93±102. van Dijk,
Teun A. 1993. Principles of critical discourse analysis. Discourse and Society 4:
249±285.
Ebden, Phillip, Arvind Bhatt, Oliver J. Carey and Brian Harrison. 1988. The bilingual
consultation. The Lancet 1988: 347.
Edwards, Alicia Betsey. 1995. The Practice of Court Interpreting. Amsterdam: John
Benjamins Publishing Company.
Elliot, Carl. 1999. A Philosophical Disease: Bioethics, Culture, and Identity. New York:
Routledge.
Erzinger, Sharry. 1991. Communication between Spanish-speaking patients and their
doctors in medical encounters. Culture, Medicine, and Psychiatry 15: 91±110. Flotow,
Luise von. 1997. Translation and Gender: Translating in the `Era of Feminism'. Ottowa:
University of Ottowa Press.
Foucault, Michel. 1963 (1973). The Birth of the Clinic: An Archaeology of Medical
Perception. New York: Vintage Books.
Foucault, Michel. 1979. The History of Sexuality, Vol. 1. London: Allen Lane. Frankel,
Richard. 1990. Talking in interviews: A dispreference for patient-initiated questions in
physician-patient encounters. In George Psathas (ed.) Interaction Com petence.
Lanham, Maryland: University Press of America, Inc. 231±262 Ginsberg, C., V. Martin,
Dennis Andrulis, Yoku Shaw-Taylor and C. McGregor. 1995. Interpretation and
Translation Services in Health Care: A Survey of U.S. Public and Private Teaching
Hospitals. Washington, D.C.: National Public Health and Hospital Institute. Goman,
Erving. 1981. Footing. In Erving Goman (ed.) Forms of Talk. Philadelphia: University of
Pennsylvania Press. 124±159.

# Blackwell Publishers Ltd. 2000

404 DAVIDSON

Gupta, Akhil and James Ferguson (eds.). 1997. Culture Power Place: Explorations in
Critical Anthropology. Durham: Duke University Press.
Haner, Linda. 1992. Translation is not enough: Interpreting in a medical setting. In Cross
Cultural Medicine ± A Decade Later [Special Issue]. The Western Journal of Medicine
157: 255±259.
Hatim, Basil. 1997. Communication across Cultures: Translation Theory and Contrastive
Text Linguistics. Exeter: University of Exeter Press.
Hatim, Basil and Ian Mason. 1989. Discourse and the Translator. London: Longman.
Hein, Norbert and Ruth Wodak. 1987. Medical interviews in internal medicine: Some
results of an empirical investigation. Text 7: 37±65.
Hewitt, William E. 1995. Court Interpretation. Williamsburg, Virginia: National Center for
State Courts.
Hymes, Dell. 1972. Models of the interaction of language and social life. In John J.
Gumperz and Dell Hymes (eds.) Directions in Sociolinguistics: The Ethnography of
Communication. New York: Holt, Rinehart and Winston, Inc. 35±71.
Juhel, Denis. 1982. Bilinguisme et traduction au Canada: role sociolinguistique du
traducteur. Quebec: International Center for Research on Bilingualism.
Kaufert, Joseph M. and William W. Koolage. 1984. Role con¯ict among `culture brokers':
The experience of native Canadian medical interpreters. Social Science and Medicine
18: 283±286.
Kleinman, Arthur. 1988. The Illness Narratives: Suering, Healing, and the Human
Condition. Harper Collins: Basic Books.
Marcos, Luis R. and Manuel Trujillo. 1984. Culture, language, and communicative
behavior: The psychiatric examination of Spanish-Americans. In Richard P. DuraÂn
(ed.) Latino Language and Communicative Behavior. New Jersey: Ablex Publishing
Co. 187±194.
Martinez, Deborah, Elizabeth A. Lenoe and Jennifer Sternback de Medina. 1985.
Language as a barrier to health care. In Lucia Elias-Olivares (ed.) Spanish Language
Use and Public Life in the United States. Berlin: Walter de Gruyter and Co. 153±164.
Mikkelson, Holly. 1998. Towards a rede®nition of the role of the court interpreter.
Interpreting 3: 21±45.
Mishler, Elliot G. 1984. The Discourse of Medicine: Dialectics of Medical Interviews. New
Jersey: Ablex Publishing Company.
Phillips, Russel S., Mary Beth Hamel, Joan M. Teno, Paul Bellamy, Steven K. Broste,
Robert M. Cali, Humberto Vidaillet, Roger B. Davis, Lawrence H. Muhlbaier, Alfred F.
Conners, Joanne Lynne and Lee Goldman. 1996. Race, resource use, and survival in
seriously ill hospitalized adults. Journal of General Internal Medicine 11: 387±396.
Putsch, Robert W. III. 1985. Cross cultural communication: The special case of
interpreters in health care. Journal of the American Medical Association 254: 3344±
3348.
Rafael, Vicente. 1993. Contracting Colonialism: Translation and Christian Conversion in
Tagalog Society under Early Spanish Rule. Durham and London: Duke University Press.
Robins, Lynne S. and Frederic M. Wolf. 1988. Confrontation and politeness strategies in
physician-patient interactions. Social Science Medicine 27: 217±221. Roy, Cynthia.
1999. Interpreting as a Discourse Process (Oxford Studies in Sociolinguistics). New
York: Oxford University Press.
Sarangi, Srikant and Stefaan Stembrouk. 1996. Language, Bureaucracy, and Social
Control. London and New York: Longman.
Scheglo, Emanuel. 1992. On talk and its institutional occasions. In Paul Drew and John #
Blackwell Publishers Ltd. 2000

INTERPRETER AS INSTITUTIONAL GATEKEEPER 405

Heritage (eds.) Talk at Work: Interaction in Institutional Settings. Cambridge:

Cambridge University Press. 101±136.
Scheppele, Kim Lane. 1989. Telling stories. Michigan Law Review 87: 2073±2098.
Shuy, Roger. 1976. The medical interview: Problems in communication. Primary Care 3:
365±386.
Snell-Hornby, Mary. 1988. Translation Studies: An Integrated Approach. Philadelphia:
John Benjamins Publishing Company.
Solis, Julia M., Gary Marks, Melinda Garcia and David Shelton. 1990. Acculturation,
access to care, and use of preventive services by Hispanics: Findings from HHANES
1982±1984. American Journal of Public Health, Volume 80 Supplement: 11±19.
Thamkins, Theresa. 1995. Culture in¯uences patients' desire to hear unfavorable
diagnosis. Asian Medical News: October 1995.
Vasquez, Carmen and Rafael A. Javier. 1991. The problem with interpreters: Commun
icating with Spanish-speaking patients. Hospital and Community Psychiatry 42: 163±
165.
Venuti, Lawrence. 1998. The Scandals of Translation: Towards an Ethics of Dierence.
London and New York: Routledge.
WadensjoÈ, Cecilia. 1998. Interpreting as Interaction (Language in social life series).
London and New York: Longman.
Waitzkin, Howard. 1983. The Second Sickness: Contradictions of Capitalist Health Care.
New York and London: The Free Press.
Waitzkin, Howard. 1991. The Politics of Medical Encounters: How Patients and Doctors
Deal with Social Problems. New Haven and London: Yale University Press. Weaver,
Charlotte Ann. 1982. Role evolution of language translators in a major medical center.
PhD Dissertation: University of California at San Francisco. West, Candace. 1984.
Routine Complications: Troubles with Talk between Doctors and Patients. Bloomington:
Indiana University Press.
West, Candace and Richard Frankel. 1991. Miscommunication in medicine. In Nikolas
Coupland, Howard Giles and John Wiemann (eds.) ``Miscommunication'' and Problem
atic Talk. Newbury Park: Sage. 166±194.
Wodak, Ruth. 1996. Disorders of Discourse. London and New York: Longman. Woloshin,
Steven, Nina A. Bickell, Lisa M. Schwartz, Francesca Gany and Gilbert Welch. 1995.
Language barriers in medicine in the Unites States. Journal of the American Medical
Association 273: 724±728.

Address correspondence to:

Brad Davidson
649 15th Avenue
San Francisco
California 94118
U.S.A.
bcdling@[Link]
# Blackwell Publishers Ltd. 2000