«JEFFREY D. ROBINSON Abstract This article deals with one form of interactional asymmetry in doctor± patient consultations, that of initiative: ...»
Asymmetry in action: Sequential resources in
the negotiation of a prescription request*
JEFFREY D. ROBINSON
This article deals with one form of interactional asymmetry in doctor±
patient consultations, that of initiative: Doctors primarily initiate actions
and solicit responses, whereas patients primarily respond to doctors' initiatives. This article argues that the variable of initiative actually contains two dimensions: speaker initiative and utterance constraint. It then reviews and critically evaluates prior accounts for these asymmetries. These accounts are almost exclusively `professional' in nature, relying upon features of the social organization of the profession of medicine, medical contexts, or institutionalized medical activities. This article argues that asymmetries of initiative can and should initially be accounted for in terms of the everyday social organization of action. The primary organizing sequential structure for action is the adjacency-pair sequence, which embodies an intersubjective set of normative standards for producing and understanding behavior. This article supports a `mundane' account of asymmetry with a conversation analytic, single-case analysis of a patient request for a renewal of a prescription.
Keywords: physician; communication; interaction; conversation analysis;
In doctor±patient consultations, there are a variety of practices of communication that are asymmetrically distributed between participants.
One important asymmetry is that of initiative: doctors primarily initiate actions and solicit responses, whereas patients primarily respond to doctors' initiatives (Frankel 1990; Linell et al. 1988; Mishler 1984;
Perakyla 1995; Todd 1993 ; West 1984).1 One primary example È È is question asking, which is signi®cantly distributed in doctors' favor (Roter et al. 1988). This has massive consequences for health care in 0165± 4888/01/0021±0019 Text 21(1/2) (2001), pp. 19±54 # Walter de Gruyter 20 Jerey D. Robinson at least three ways. First, question asking by doctors is negatively correlated with patients' adherence to medical advice (Hall et al. 1988).
Second, patterns of communication characterized by high frequencies of question asking by doctors are negatively correlated with patients' satisfaction (Roter et al. 1997). Third, patients' questions are a major vehicle for gaining information from doctors, which is positively correlated with patients' satisfaction, adherence, and recall/understanding (Hall et al. 1988). Of course, question asking is merely one incarnation of the asymmetry of initiative. This asymmetry is generally conducive to doctors' control of the ¯ow of information and topics, which is conducive to the suppression of patients' lifeworld experiences in favor of doctors' biomedical experiences and the reduction of information relevant to diagnosis and treatment (Beckman and Frankel 1984; Beckman et al.
1985; Fisher 1991; Larsson et al. 1987; Lipkin et al. 1995; Marvel et al. 1999; McWhinney 1981, 1989; Mishler 1984; Sankar 1986; Todd 1993 , 1989).
Because of its relevance to health-care outcomes, researchers have attempted to account for the initiative asymmetry. However, such accounts must be grounded in experiences that are lived, and oriented to as relevant, by doctors and patients. In both mundane and medical interaction, interactants conduct themselves primarily in terms of action (Drew and Heritage 1992; Scheglo 1995a). Although numerous researchers have coded for doctors' and patients' `questions' and `answers', these labels are not necessarily adequate descriptions of actions because interrogatively formatted utterances and their responses frequently perform a variety of actions other than, or in addition to, gathering and providing information (Atkinson and Drew 1979; Frankel 1990; Heritage and Roth 1995; Scheglo 1984, 1995a). In primary-care consultations, one common course of action is patients' requests for non-diagnostic service(s), such as requests for prescriptions. These requestsÐwhich are frequently interactionally extended and complex, containing a myriad of component actions, such as taking, advocating, and resisting positions regarding the decision, and soliciting and providing information in the service of making the decisionÐare socially organized and provide analytically fertile ground for accounting for the initiative asymmetry.
This article does four things. First, it argues that prior operationalizations of `initiative' have con¯ated two variables: speaker initiative and utterance constraint. Second, it reviews and critically evaluates two
types of accounts that prior research has oered for these asymmetries:
accounts that are exogenous and endogenous to interaction. These accounts are almost exclusively `professional' in nature, relying upon Asymmetry in action 21 features of the social organization of the profession of medicine, medical contexts, or institutionalized medical activities. Third, it argues that these asymmetries should initially be accounted for in terms of the interaction order sui generis. Speci®cally, it argues for a mundane, or lay, account in terms of the everyday social organization of action, focusing on the adjacency-pair sequence (Sacks 1992b; Scheglo 1968; Scheglo and Sacks 1973). Fourth, it supports this argument with a conversation analytic, single-case analysis of a patient request for a renewal of a prescription, throughout which implications are drawn for research on asymmetry.
A clari®cation of the initiative asymmetry Although the initiative asymmetry is sometimes discussed as a single phenomenon, it actually has multiple aspects. Because all utterances are simultaneously context sensitive and context renewing (Heritage 1984b),
there are at least two dimensions of initiative concerning an utterance:
1. the range of initiative that speakers can be said to have when producing it;
the range of initiative it provides for potential next speakers.2 2.
Along these lines, the production of an utterance can be grossly categorized along two dimensions: speaker initiative and utterance constraint. To elaborate, it is necessary to introduce the concept of the adjacency-pair sequence (Sacks 1992b; Scheglo 1968; Scheglo and Sacks 1973). In its basic form, the adjacency-pair sequence is composed of two parts, a ®rst-pair part and a second-pair part, each produced by dierent speakers. The ®rst-pair part constitutes an initiating action (e.g., a request) that normatively obligates the selected next speaker to produce a relevant and responsive second-pair part (e.g., a granting). Actions are pair typed, meaning that ®rst-pair parts make relevant a ®nite range of second-pair parts (e.g., a request is relevantly responded to with a decision to either grant or deny the request, not with a greeting, an assessment, etc.).3 The normative obligations of a ®rst-pair part establishes an inferential framework for action such that the selected next speaker is accountable for immediately producing a relevant second-pair part (with respect to accountability, see Heritage 1984b). In terms of the dimension of utterance constraint, utterances can either normatively obligate a response from a recipient (as ®rst-pair parts obligate second-pair parts) or not obligate a response.
In terms of speaker initiative, utterances can be either normatively obligated by a prior utterance (as second-pair parts are obligated by ®rst-pair parts) or not normatively obligatedÐthat is, volunteered by speakers.
22 Jerey D. Robinson Prior research has sometimes con¯ated asymmetries of speaker initiative and utterance constraint. The most common operationalization of initiative has been in terms of adjacency-pair sequence structure (i.e., ®rst- and second-pair parts; Frankel 1990; Linell et al. 1988; Todd 1993 ; West 1984). Here, the ®ndings are that doctors' turns are primarily ®rst-pair parts (e.g., questions) whereas patients' turns are primarily second-pair parts (e.g., answers). However, some researchers have simultaneously operationalized initiative positionally. For example, Frankel (1990) and West (1984) also coded for `®rst-positioned' utterances that are, in my terms of speaker initiative, volunteered, but that, in my terms of utterance constraint, do not necessarily obligate a response.4 Here, the ®ndings are that after patients provide complete responses to doctors' ®rst-pair parts, if patients continue to volunteer additional talk, they often format that talk as additional responsive components to doctors' prior ®rst-pair parts, rather than as stand-alone actions that may or may not obligate a response (Frankel 1990; Gill 1998; Gill and Maynard to appear; Linell et al. 1988). In sum, the general notion of initiative embodies at least two analytically distinct concepts, speaker initiative and utterance constraint, each of which have dierent implications for conceptions of asymmetry.
Existing accounts for asymmetries of speaker initiative and utterance constraint: Exogenous and endogenous Because of their relevance to health-care outcomes, researchers have attempted to account for asymmetries of speaker initiative and utterance constraint. One possible account is in terms of factors that are exogenous to interaction. For example, some researchers have theorized that the medical profession contains institutionalized power roles of dominance and subordination for doctors and patients, respectively (Freidson 1970a, 1970b; Navarro 1976; Parsons 195l, 1975; Starr 1982; Waitzkin and Waterman 1974). These asymmetries might be conceptualized as a byproduct of doctors' and patients' subscription to these roles. A similar account might been given in terms of a variety of sociopolitical structures that are associated with dominance, such as class (Cartwright 1967; Waitzkin 1985), ethnicity (for review, see Roter and Hall 1992), gender (Pendleton and Bochner 1980), and socioeconomic status (Sleath et al. 1997).
There are at least two reasons why these exogenous accounts are insucient. First, there is no necessary relationship between these asymmetries and power or dominance.5 Researchers have coded and counted ®rst- and second-pair parts, as well as ®rst-positioned utterances, largely Asymmetry in action 23 without characterizing the actions that doctors and patients are accomplishing with those utterances. Although there is a distinction between the formal structure of the adjacency pair and the action(s) being accomplished through it, this is a distinction for analysts, not participants, who orient to ®rst- and second-pair parts in terms of the action(s) they accomplish. Because the actions accomplished through ®rst- and second-pair parts can either maintain or challenge traditional power/ dominance relationships, it is con¯ationary to conceptualize dominance in terms of their aggregated distributions among doctors and patients.6 Second, a large body of research has demonstrated that actual medical interaction does not consistently embody, and sometimes contradicts, theoretical, social-structural relationships as they relate to asymmetrical distributions of communication practices (Anspach 1993; Becker et al.
1961; Beisecker and Beisecker 1990; Bloor 1976; Emerson 1994 ;
Strong 1979; Sudnow 1967; ten Have 1991; West 1976). Research has since demonstrated that interaction has its own, independent order of social organization (Goman 1983; Scheglo 1987b) and that interactional asymmetries are collaboratively accomplished in and through interaction (Cicourel 1973; Heath 1992; Linell and Luckman 1991;
Maynard 1991). While not ¯atly rejecting exogenous accounts for interactional asymmetries, this research argues that any account must initially be sought in factors that are endogenous to interaction (Drew and Heritage 1992; Maynard 1991; Mishler 1984).
Prior research has oered three types of endogenous accounts: the speech exchange account, the chaining rule account, and the professional activity account. According to the speech-exchange account, doctors and patients organize interaction according to a formal, professional speech exchange (i.e., turn-taking) system in which patients' ®rst-pair parts and other ®rst-positioned actions are normatively inappropriate and avoided (Frankel 1990; West 1984).7 Some have gone so far as to argue that this turn-taking system preallocates turn order and turn type into a three-part sequential structure, including (1) doctors' initiations of actions; (2) patients' responses; and (3) doctors' third-turn responses, such as (dis) agreements with, and assessments of, patients' responses (Fisher 1984; Mishler 1984; Todd 1993 ). According to this account, asymmetries of speaker initiative and utterance constraint are features of the interactional structure of consultations themselves as achieved forms of professional social order.8 However, this account does not hold up. Despite the distributional fact that these asymmetries exist, there is evidence that doctors and patients structure consultations, unproblematically and without sanction, according to the turntaking rules of mundane conversation (Anderson 1979; Perakyla 1995;
È È 24 Jerey D. Robinson ten Have 1991). Furthermore, there is evidence that, even during periods of asymmetry, doctors and patients display that they are oriented to the locally managed turn-taking rules of mundane conversation (PerakylaÈ È 1995). This evidence rejects the claim that doctor±patient consultations embody a distinctive turn-taking system (Heritage 1998).
According to the chaining rule account, which Frankel (1990) borrowed from Churchill's (1978) interpretation of Sacks (1992d), the rules for turn taking in mundane conversation (Sacks et al. 1974) provide questioners in two-party conversations with a priveledged opportunity to continue questioning. As Sacks articulated it, A person who has asked a question [has] a `reserved right to talk again,' after the one to whom he has addressed the question speaks. And, in using the reserved right, he can ask [another] question (Sacks et al. 1974: 264).
Although the chaining rule may account for some asymmetry, it is, at best, only a partial account. Sacks's point was not that the turn slot for answering is an iron cage. Indeed, i. speakers have a variety of practices for securing multi-unit responses (Goodwin 1996; Jeerson 1986; Scheglo 1982, 1996b);