Epistemics

Epistemics JACK SIDNELL University of Toronto, Canada

Current research into epistemics, a concern within conversation analysis, considers the various ways in which participants in interaction design their contributions, and understand the contributions of others, in relation to a distribution of knowledge that is assumed to preexist. Following Heritage (2012a, 2012b), this article suggests that a wide range of phenomena within this general domain of conversational organization can be understood through the central ideas of epistemic status and epistemic stance. Epistemic status refers to a distribution of knowledge among participants that is assumed to preexist and to the rights and entitlements that flow from it. For instance, a person is typically understood to know his or her own experience better that anyone else. In consequence a person has special and primary rights to talk about what s/he has said, done, felt, perceived, and so on. Epistemic entitlements of this kind are not, however, based on access alone and extend beyond one’s own experience to those to whom one is related. Thus a mother is treated as having special rights to talk about her child, even if another person, such as a day-care worker, has strictly equivalent or even better (e.g., more recent, more extensive, etc.) access. Because the entitlements that constitute epistemic status flow from social relations as much as they do from access, persons find themselves operating within a rich web of sometimes conflicting rights and obligations to know. Epistemic stance on the other hand refers to the multitude of ways in which a turn at talk can convey, through the details of its design, the differential knowledge entitlements of speaker, recipient, and other participants. So, for instance, epistemic stance is expressed in the various grammatical formats used to construct yes/no questions in English, from the relatively certain declarative (declaration + tag) format (“You’ve been down here before, haven’t you?”) to the relatively tentative auxiliary inversion format (“Are they right up to the houses?”) and so on. A wide range of other linguistic resources are used to convey epistemic stance: so-called evidential markers, “parenthetical verbs” such as “I think,” intonation contours, and so on (for a review, see Sidnell, 2012).

Background The term “epistemics” is derived from the Greek word epist¯em¯e meaning “knowledge, understanding.” In philosophy, “epistemology” concerns the nature, sources, and limits of human knowledge and justification. The standard analysis of knowledge in philosophy as “true, justified, belief” is typically attributed to Plato. On this view, knowledge is propositional. Many philosophers have questioned this. Gilbert Ryle for instance argued for the importance of distinguishing knowing how (e.g., knowing how to count) The International Encyclopedia of Language and Social Interaction, First Edition. Karen Tracy (General Editor), Cornelia Ilie and Todd Sandel (Associate Editors). © 2015 John Wiley & Sons, Inc. Published 2015 by John Wiley & Sons, Inc. DOI: 10.1002/9781118611463/wbielsi143

2

EP I S T E M I C S

and knowing that (knowing that 2 + 2 = 4). According to Ryle, knowing how always underlies any knowing that. Another important distinction, made by Bernard Russell, is that between knowledge by description and knowledge by acquaintance. Knowledge by description is knowledge acquired through another person’s testimony. Knowledge by acquaintance is knowledge acquired at first hand, through experience or perception. Epistemological concerns figured centrally in the development of phenomenology, especially in the work of Husserl and Schütz. In his writings, Schütz drew on the phenomenological method to elaborate an early account of the social distribution of knowledge and its consequences for self–other relations (i.e., for intersubjectivity). Through its influence on Mannheim and Garfinkel, phenomenology also played an important role in the articulation of what became known as “the sociology of knowledge.” The beginnings of the sociology of knowledge can be traced to the writings of Durkheim. In his work with Mauss, which was translated under the title Primitive Classification (Durkheim & Mauss, 1963), Durkheim argued that forms of social organization shape local modes of categorization. Here Durkheim introduces the idea of “collective representation”: symbols or images that represent the ideas, beliefs, and values elaborated by a group. Because they are the product of collective, ritual activity these representations (words, slogans, material items) cannot be reduced to individual psychology. They are rather the sui generis products of a society. According to Durkheim, these collective representations serve both as norms that constrain individual behavior and as the means by which persons make sense of the world and find order in it. To summarize, according to Durkheim and Mauss, primitive thinking is shaped by local forms of social organization and ritual activity. Much thinking is done through the vehicle of collective representations and thus is fundamentally socially organized. Under the influence of Marx and Engels, Karl Mannheim developed an alternative view of the sociology of knowledge. Marx and Engels (1845/1966) had proposed, in various ways, that consciousness was a product of material and social conditions; they wrote for instance that “the production of ideas, of conceptions, of consciousness, is at first directly interwoven with the material activity and the material intercourse of men, the language of real life” (1845/1966, p. 47). According to Mannheim, knowledge and beliefs are the products of sociopolitical forces, a position that tends to reduce epistemological questions to ideological ones. Despite differences in this respect, there are clear parallels with the account developed by Durkheim. In both cases, the primary concern is to elucidate the “external” forces that shape what people (think that they) know. That is to say, beliefs (aka knowledge) are examined as constructions that are supported by and at the same time serve to support existing social arrangements that are external to them. An alternative approach, one that animates the work of Schütz (1946) and a number of other sociologists and anthropologists, views the real or assumed distribution of knowledge as internal to and constitutive of social structure and relations. On this account, differential knowledge is a key aspect of basic social distinctions. Two groups within a society (e.g., different jatis in an Indian village) may be distinguished in large part through their knowledge of ritual procedures. And, of course, roles such as that of nurse or doctor, student or teacher may be distinguished in part by an assumed, normatively prescribed distribution of knowledge. Whereas classic work

EP I S T E M I C S

3

in the sociology of knowledge has been primarily concerned with the external social and political forces that shape knowledge, this alternative view treats knowledge as a constitutive feature of social life and social relations. In what is perhaps the culmination of this view, Foucault proclaimed that “knowledge is power” and argued that it was in and through their articulation by “regimes of knowledge” that persons are constituted as subjects (e.g., Foucault, 1980). Another key set of contributions come from work in social psychology and linguistics. In psychology, Clark (1996) has proposed that assumptions about common ground—either culturally shared knowledge or contextually shared presuppositions—play an important role in the interpretation of talk and other conduct. Similarly, a key notion in linguistic and philosophical pragmatics is that any given utterance semantically encodes and pragmatically presupposes information. In linguistics, a number of scholars such as Labov, Bolinger, Kamio have drawn attention to the way assumptions about the distribution of knowledge between participants guide the understanding and use of a range of utterance types. Finally, work in linguistic and philosophical pragmatics has attempted to clarify the relationship between lexical semantics and contextual inferences through analyses of presupposition, entailment, implicature, relevance, and so on. A final component of the background to critical analysis (CA) studies is work in linguistics on evidentiality and related topics. Linguists use the term “evidentiality” to refer to a grammatical category, the exponents of which are typically assumed to convey the source of knowledge or evidence for a given statement. In fact the semantics of the various elements referred to as evidentials is a good deal more complicated than this suggests. In actual usage evidentials may be more associated with the degree of speaker certainty than with the actual source of knowledge. While linguists typically reserve the term “evidential” for a fully grammaticalized and obligatory marking, in many languages similar meanings are conveyed by grammatically optional (even if pragmatically expected) lexical items such as (in English) “seems” or “appears” or “I think” (see Sidnell, 2012). And, of course, the degree of speaker certainty may be conveyed by quite different means, such as by selecting an interrogative (as opposed to a declarative) format.

The conversation-analytic approach Current work on epistemics within CA is grounded in pioneering work carried out in this area by Sacks, Pomerantz, and Schegloff. In his lectures on stories, Sacks (1995) formulated an early version of the recipient design principle through the maxim “Don’t tell others what they already know” and went on to show that, in telling and listening to stories, participants hold one another accountable in terms of what each is warranted or obligated to know. Terasaki’s (1976/2004) early study of preannouncement sequences developed this idea, showing that speakers engage in such activities in order to determine whether recipients already know the news or not. If they do not already know, the telling can proceed without risk of violating the principle of recipient design; if they already know, the speaker can abort the projected telling. Similarly, in their work on person reference, Sacks and Schegloff (1979) and Sacks (1995) noted the importance of a speaker’s assumptions about what the recipient knows.

4

EP I S T E M I C S

That is, in making reference to some specific person through use of the name “John,” the speaker assumes that the recipient knows this person, knows him by this name, and knows that the speaker knows him by this name. The basic idea here is that referring to a person through “John” or some other recognitional item involves a set of assumptions about what knowledge is shared between speaker and recipient. Person reference here merely illustrates a much more general phenomenon, since any action in conversation rests upon a foundation of mutual knowledge. These studies focused on the important role that assumptions about shared knowledge play in the organization of interaction; they further the work that participants do to determine the degree to which knowledge is indeed shared or not. In her study of actions implemented through talking about something to which one has limited access—what has become known as “my-side tellings”—Pomerantz (1980) drew attention to the interactional consequences of the differential distribution of knowledge. According to Pomerantz, my-side tellings serve as a technique for eliciting information without having to ask a direct (or any) question. Actions are accountable. This means that a person asking a question should be entitled to ask it and, moreover, should have a good reason for doing so. Responses to questions such as “Why do you want to know?” or “Why do you ask?”—or even replies like “It’s none of your business”—hint at this accountability of actions, as do reports about a person being a “gossip,” “a nosey parker,” or “a busy-body.” With this is mind, we can see how the practice Pomerantz describes might have important interactional uses, since with its help a person may elicit information about some specific state of affairs without being accountable for having asked a question. Consider the following case, in which, by saying “Yer line’s been busy” and thereby conveying “limited access” to some set of events that the recipient knows about, Emma is “soliciting” information from Nancy. The next turn here seems to make the case, as Nancy provides information about how it was that her phone was busy. (1) Pomerantz (1980, p. 189): NB:II:2:R 01 02 03 04 05 06 07 08 09 10 11

NANCY: EMMA: NANCY: EMMA: -> NANCY:

EMMA: ->

Hel̂lo:, .hh HI::. (.) Oh: ‘i::: ‘ow a:re you Emmah: FI:NE yer LINE’S BEEN BUSY. Yea:h (.) my u.-fuhh h-.hhhh my fa:ther’s wife ca:lled me,h .hhh So when she ca:lls me::,h .hh I always talk fer a lo:ng ti:me cz she c’n afford it’n I ca:n’t.hhh [hhh] ∘ huh∘ ] [OH:]:::::]:=

In the course of her analysis Pomerantz draws upon a distinction introduced by Sacks between “direct”—type 1—and “indirect” or “derived”—type 2—knowledge. Type 1 includes knowledge about one’s own experience, thoughts, and feelings, as well as about one’s own activities and history. Pomerantz (1980, pp. 187–188) describes the distinction in the following way:

EP I S T E M I C S

5

Type 1 knowables are those that subject-actors as subject actors have rights and obligations to know. For example, one’s name, what one is doing, and so on are assumed to be available to a competent subject-actor. Type 2 knowables are those that subject-actors are assumed to have access to by virtue of the knowings being occasioned. Where your friend is, what she or he did yesterday, and the like are accountably available by virtue of the subject-actor’s having been told, having figured it out, having seen the friend, and so on.

An important point here is that the distinction is actually more about rights and obligations to know than it is about access per se. That is, what is being described is a moral and normative distribution of rights to knowledge. As already noted, Heritage (2012a) uses the phrase “epistemic status” to describe this presumed, normative distribution of knowledge. He writes: relative access to particular epistemic domains is treated as a more or less settled matter in the large bulk of ordinary interaction. Outside of very specialized contexts such as psychoanalysis, the thoughts, experiences, hopes and expectations of individuals are treated as theirs to know and describe … Persons are generally treated as knowing more about their relatives, friends, pets, jobs and hobbies than others, and indeed may labor under an obligation to do so. Moreover there are socially sanctioned ways to adjudicate epistemic disputes in which, for example, recent experience is privileged over less recent, or external expertise and epistemic authority are permitted to trump the judgments of amateurs … For these reasons, notwithstanding the vagaries of its status as a social construction, it is helpful to think of epistemic status relative to a domain as for the most part a real, enduring, and generally agreed-upon state of affairs. (Heritage, 2012a, p. 6)

In another context Heritage (2012b) suggests that epistemic status can be conceptualized as relative positioning along an epistemic gradient. The notion of epistemic status draws together the ideas of Labov, Pomerantz and Kamio. It begins from the notion that any two speakers, A and B, each have their own territories of information, and that any specific element of knowledge can fall into both of them, but often to different degrees. Relative epistemic access to a domain will be stratified between A and B such that they occupy different positions on an epistemic gradient (more knowledgeable [K+] or less knowledgeable [K-]), which itself may vary in slope from shallow to deep … We will refer to this relative positioning as epistemic status and stipulate that it involves the parties’ joint recognition of their comparative access, knowledgeability, and rights relative to some domain of knowledge as a matter of more or less established fact. This relative positioning may vary from nearly absolute inequality to nearly absolute equality. (Heritage, 2012b, p. 376)

The backdrop of ordinary interaction then includes a set of pervasive assumptions about who (should) know what and how well or to what extent they (should) know it relative to others with whom they interact. Whereas the notion of epistemic status points to a relatively stable and enduring presumed distribution of knowledge and of its associated entitlements, epistemic stance, by contrast, concerns the moment-by-moment expression of epistemic relations through the design of turns at talk. Epistemic stance is conveyed through a very wide range of features (see Sidnell, 2012 for an overview) but, perhaps most importantly, in English through grammatical design. So for instance a proposition may be conveyed on the one hand through a bald assertion (“You’ve been down here”), through a tagged assertion (“You’ve been down here, haven’t you?”), or through an interrogative (“Have you been down here?”). In each of these the same content is expressed, but the epistemic

6

EP I S T E M I C S

stance encoded in the grammar is different. In English, epistemic stance is also frequently expressed by grammatically optional lexical items such as “perhaps,” “apparently,” “seems,” and so on. Such items suggest that a speaker has only indirect or inferential grounds for whatever is being proposed. In many languages an item that specifies the source of knowledge is required and a sentence without such “evidential” marking may be found ungrammatical.

Epistemics in action ascription A basic problem for the recipient of an utterance is to determine what action(s) it is meant to accomplish. Heritage (2012a) suggests that epistemics plays an important role in the solution to this problem. Specifically, a recipient will often draw on assumptions about who knows what (epistemic status) in deciding whether a given utterance is asking or telling. So, for instance, if, during a telephone conversation, the speaker says “It’s raining” (with intonation that does not disambiguate between assertion and question), she may be understood as asking a question if she just woke up and is inside the house while speaking to her friend, who is in the park. If, however, the speaker is in the park while the recipient is still inside, the utterance is more likely to be understood as an assertion. While the basic idea that a declaratively formatted utterance is typically treated as a question (request for confirmation) was proposed some time ago, the broader implications for our understanding of interaction have been made explicit in recent work by John Heritage. Heritage (2012a) proposes that, where epistemic stance and status are congruent, a recipient will typically read the intended action from grammatical form, such that a declarative utterance will be understood as asserting the proposition and so on. However, where the stance and status are noncongruent, this may license an inference that the speaker is requesting information (i.e., is questioning) in spite of the fact that the utterance is formatted as a declarative. Epistemic status, it is suggested, takes precedence over a turn’s grammatically displayed epistemic stance in terms of what action a turn is understood to be implementing. Heritage proposes: “Insofar as asserting or requesting information is a fundamental underlying feature of many classes of social action, consideration of the (relative) epistemic statuses of the speaker and hearer are [sic] a fundamental and unavoidable element in the construction of social action” (Heritage, 2012a, p. 1). Roughly speaking, Heritage shows that a recipient’s assumptions about what the speaker does or does not know are decisive in determining whether a declaratively formatted utterance is heard as telling or asking. Robinson (2013) discusses a similar set of issues involved in determining what “repair-related” action a speaker is doing in initiating repair with a partial questioning repeat. According to Robinson, in order for a recipient to understand the particular social action being accomplished by a partial questioning repeat, the recipient must determine how much the producer knows about the repeated item in context, that is, how thoroughly, accurately, and/or authoritatively the producer understands the meaning of the repeated item in the context of the unit of talk that contains the putative trouble. (Robinson, 2013, p. 265)

7

EP I S T E M I C S Table 1 Epistemics and action formation.

Turn design feature

K+ epistemic status (within speaker’s epistemic domain) Action interpretation

K– epistemic status (not within speaker’s epistemic domain)

(Given the “known in common” epistemic status of speaker and recipient relative to the targeted state of affairs) Declarative syntax Declarative syntax with final rising intonation Tag questions Negative interrogative syntax Interrogative syntax

Informing Continuing

Declarative/Bevent question Questioning

Mobilizing support for an assertion Asserting information Preinforming question Known-answer question Rhetorical question

Seeking confirmation Requesting information Request for information

Source: Adapted from Heritage, 2012a, p. 24. Epistemics in action: Action formation and territories of knowledge, J. Heritage, Research on Language and Social Interaction, 2012, Taylor and Francis, reprinted by permission of the publisher (Taylor & Francis Ltd, http://www.tandf.co.uk/journals).

On the one hand, if the recipient figures that the producer has knowledge of the repeated item, the recipient will understand the partial questioning repeat as implementing a particular class of repair-related [K+] actions, which index a speaker’s (broadly speaking) “disagreement” with the repeated item. Here “disagreement” can range from pro forma disagreement, which indicates ritualized disbelief or surprise, to “serious” disagreement, which challenges the relevance, appropriateness, or accuracy of the repeated item. If, on the other hand, the recipient figures that the speaker of the partial questioning repeat is not knowledgeable with respect to the repeated item, then the recipient will likely understand it to be implementing a different set of repair-related actions, which Robinson refers to as [K–] actions: These convey their speakers’ lack of understanding or problematic hearing of the repeated item. The studies by Heritage and Robinson show that a recipient’s understanding of a declaratively formatted turn-constructional unit (TCU) or of a partial questioning repeat is shaped by assumptions about the background distribution of knowledge and, specifically, about what the speaker does or does not know. These studies and others discussed below employ a simple [K+] and [K–] notation to represent these various possibilities. Heritage (2012a, 2012b) summarizes in Table 1 the role that epistemic status plays in action ascription for a variety of grammatical formats.

Repair and epistemic entitlements In human interaction there is always a possibility that troubles will arise in speaking, hearing, and understanding (Schegloff, Jefferson, & Sacks, 1977). An organized set of practices of repair constitutes a natural, interactive system by which such troubles may be addressed at or near the point where they occur and may be resolved more or less immediately, before there is intolerable divergence among the participants’

8

EP I S T E M I C S

intersubjective understandings of what is going on in the interaction. In their pioneering work on repair, Schegloff and colleagues argued that a range of phenomena that had been considered in isolation up to that point were in fact bound together into a single system through a preference for self-correction. Their analysis showed that the speaker of a trouble source not only has the first opportunity to repair it but is also accorded greater rights to do so within the organization of norms that oversee the practices for dealing with trouble. The rights that a speaker has to produce and to repair his or her own talk are conferred by the turn-taking system, which, as Schegloff and colleagues demonstrated, grants to speakers the right (and the obligation) to produce a single TCU through to its first point of possible completion. Recent work indicates that this preference for self-correction is modulated by epistemic considerations. Bolden (2013), for instance, shows that, where two persons are knowledgeable about some state of affairs, either one of them may respond to a repair initiation, regardless of which one produced the trouble source. In a case such as this, then, epistemic status trumps the preference for self-correction. Sidnell and Barnes (2013) consider various ways in which a description may be questioned, challenged, or corrected in a subsequent turn. The analysis begins by making a distinction based upon whose epistemic domain the description pertains to. For instance, if a first speaker says “when we were talkin’ about getting rid of all of your accomplishment,” what is being described as “your accomplishment” belongs to the recipient’s epistemic domain. If, on the other hand, a first speaker says “I have enjoyed this business an’ I developed some knack for it,” what is described as “developing a knack” is squarely in the initial speaker’s epistemic domain. Taking the perspective of the next speaker, Sidnell and Barnes conceptualize this as a distinction between “mydomain” and “your-domain” descriptions and show that participants are normatively entitled to replace a description in the next turn if it falls within their own epistemic domain, whereas they are only entitled to “identify” a problematic description that falls in the first speaker’s epistemic domain. The examples below illustrate this rule: (2) Sidnell & Barnes (2013, p. 331) 01 02 03 04 05 06 07 08

ADULT: A: –> C: –>

A: C: A: C:

Guys too loud. She po: :ked it. I ta:pped it. (0.2) Well you knocked it over. No I didn’t. Yes you did. Oh whatever.

Here the children are playing with blocks. When the structure falls, A screams and, at line 1, the supervising adult complains: “Guys too loud.” This occasions an excuse from A, who assigns responsibility to C by saying “She po::ked it”—with reference to the structure she is building. Notice, then, that the description concerns C’s action and thus falls squarely within C’s epistemic domain. In the next turn C replaces “po::ked” with “ta:pped.”

EP I S T E M I C S

9

(3) Sidnell & Barnes (2013, p. 327) 09 HUNTER: 10 -> 11 12 13 I: -> 14 15 16 HH:

I have enjoyed this business an’ I developed some knack for it. and uh .h I’ve stayed with it fer a long time. Some knack for it. I think some would say that’s an understatement. (0.2) Well thank you.

In (3), however, when Hunter suggests “I developed some knack for it,” the next speaker does not replace this description but merely identifies it as problematic by repeating it and then characterizing it as an “understatement.” These studies show, then, that the preference for self-correction, which organizes the domain of repair practices in conversational interaction, is modulated by epistemic matters. The preference for self-correction derives from entitlements conferred by the turn-taking system, specifically the entitlement to “speak for oneself.” In these studies of repair we see how these rights and entitlements conferred to the turn-taking system operate alongside another set of rights and entitlements, conferred by the normative organization of knowledge—that is, by epistemics.

Epistemics, agreement, and confirmation Much of the recent CA literature on epistemics in interaction has focused on assessment sequences. An assessment is an utterance that expresses its speaker’s positive or negative stance toward some person or object talked about, for example “It’s a fantastic film,” or “That is so gross.” Utterances like these typically (though not invariably) make agreement the relevant next action, so that any other action will be heard as tantamount to disagreement (that is, as withholding agreement). In an influential paper, Heritage and Raymond (2005) note that to agree with an assessment presupposes some access to whatever is being assessed. For instance, in the following case the participants (J and L) are talking about the weather, which is equally available to both. (4) Heritage & Raymond (2005, p. 17) 01 02

J: L:

T’s tsuh beautiful day out isn’t it? Yeh it’s jus’ gorgeous…

Notice that, in (4), J designs her assessment as a question, to invite L’s confirmation; in this way J displays her own sense that the weather is something more or less equally available to both participants. Because they presuppose joint or shared—though not necessarily equivalent— access, sequences of assessment and agreement routinely involve participants who may rank the immediacy or quality of their access. In assessment sequences of this kind,

10

EP I S T E M I C S

agreement can easily lead into negotiation of “who knows better” or “who knows best.” Even where the participants agree, then, they can frequently be seen to compete over the matter of who knows more, or better, about whatever it is being evaluated. This competition takes place within an organized structure of positions. Thus first-position assessments carry an implicit claim to primacy, and second assessors are vulnerable to being heard as merely going along with a first. It is therefore not surprising that most of the epistemic practices described in relation to assessments have to do with second speakers “getting out from under” the implications of second position or with first speakers downgrading claims that might otherwise accrue to first-positioned assessments. Consider the fifth example, below. (5) Heritage & Raymond (2005, p. 18): SBL 2-1-8:5 01 02 03 04 05 06 07

Bea: Nor: -> Bea:

Nor:

hh hhh We:ll,h I wz gla:d she c’d come too las’ni:ght= =Sh[e seems such a n]ice little [l a dy] [dAwf’l]ly [(since you keh) ] nice l*i’l pers’n. t hhhh hhh We:ll, I[: j’s ] [I thin]k evryone enjoyed jus ...

Heritage and Raymond note that Norma had met the person being assessed in line 2—a long-time friend of Bea’s—the night before for the first time. This “asymmetry in their experience is indexed in Norma’s evidentially downgraded assessment” (she seems such a nice little lady) and in “Bea’s declaratively asserted agreement.” The authors conclude: “By downgrading her claimed access, Norma manages her initial assessment so as to defeat any epistemic priority that might have been inferred from its sheer first positioning” (p. 18). On the other side of things, Heritage and Raymond discuss a range of practices that second speakers use when they want to upgrade their claimed access. One practice involves formatting the response as a [confirmation + agreement token]. Another practice the authors consider is “oh”-prefacing. Heritage (2002) has shown that “oh” conveys a “change of state.” In the case of oh-prefaced responses to assessments, “oh” conveys that the first assessment has “occasioned a review, recollection and renewal of the speaker’s previous experience and judgment, and that it is this review that forms the basis for the second assessment” (p. 201). “Oh”-prefacing thus indexes the speaker’s prior and previous experience or position and thus provides a “systematic way of claiming that a speaker has independent access to, and already holds a position regarding the referent” (p. 201). The second assessment thus shows itself to be independent of its “second position” and based on quite independent grounds. Raymond and Heritage (2006, p. 684) note that “epistemic rights to assess are not solely (or even mainly) distributed on the basis of physical access to a referent state of affairs; they are socially distributed.” This can be seen in cases where two quite different bases for knowing are concurrently relevant in an interaction. In talking about a particular child, for instance, one participant may have epistemic rights on the basis of the fact that she is the child’s grandmother, while the other may have rights

EP I S T E M I C S

11

on the basis of the fact that she has seen the child more recently, during a visit to her house. Stivers (2005) describes a number of practices that speakers use for confirming prior talk—an action that can be thought of as constituting a special case of agreeing. One practice involves repeating what another has said in an immediately previous turn. In (6) three children are playing with blocks, and in line 1 G announces: “I’m gonna see if this will work.” She then places the marble in the maze of blocks and, when it comes out the other end, announces “okay it does.” However, the marble then begins to roll back into the maze, occasioning G’s modification with “well almost” and T’s “No it doesn’t.” J then confirms what G has said by repeating her words and appending “yeah.” (6) Sidnell (2012, p. 309) 01 G: 02 03 04 05 -> 06 T: 07 J: =>

I’m gonna see if this will work. (1.0) Okay it does_ (0.2) W[ell almost. [No it doesn’t. Well almost yea:h.

Notice that in this case G’s turn at line 5 did not invite confirmation or disconfirmation. So we can see that confirmation is a special way of agreeing and that it can occur both in contexts where it was invited by a prior turn at talk and in contexts where it was not. Stivers (2005) argues that such uninvited confirmations carry with them an epistemic challenge. Specifically, confirming what another has said where this has not been invited is to challenge the implicit claim to epistemic priority that accrues to a first-position assertion.

Conclusion: Conceptualizing the distribution of knowledge In the studies reviewed here we can see the central role that epistemic matters play in the organization of interaction. Specifically, we can see that participants in interaction display a ubiquitous orientation to the matter of who knows what and how well they know it. For the most part, research in this area has adopted a simple notation of [K+] and [K–] to represent epistemic differences between participants and, along with this, a view of the social distribution of knowledge as a simple, one-dimensional gradient. For many, perhaps most environments, this simple conceptual model seems warranted, since participants themselves organize their conduct in relation to such a heuristic. However, it is also possible that, in at least some situations, participants are working with a more complex, multidimensional sense of epistemic territories. In “The well-informed citizen: An essay on the social distribution of knowledge,” Schütz (1946) suggested that,

12

EP I S T E M I C S

if we had to draw a map depicting the distribution of knowledge in society … it would not resemble a political map showing the various countries with their well-established frontiers but rather a topographical map representing the shape of a mountain range in the customary way by contour lines connecting points of equal altitude. Peaks and valleys, foothills and slopes, are spread over the map in infinitely diversified configurations. (Schütz, 1946, p. 469)

There are cases in interaction that seem to presuppose a more complex sense of the distribution, along the lines of what Schütz suggests here. For instance, in a situation where a masseuse and a client are talking about the tightness of the client’s shoulder muscles, both participants have direct, unmediated access to the worldly phenomenon being discussed (the muscles in the client’s shoulders), but access that is fundamentally different in kind. Here the notion of a gradient, with participants positioned as [K+] and [K–], provides little traction. Instead we need to recognize the different standpoints for which it is possible to be knowledgeable about something. Because epistemic entitlements do not flow directly from “access” but rather are shaped by social norms and expectations, there are many cases in which participants’ knowledge about ostensibly the same state of affairs is supported by quite different authorities. So, for instance, a doctor may have expert knowledge of a medical condition (and associated knowledge about what causes it, what is the likely outcome, etc.), while a patient has direct experience of the physical effects and symptoms of the condition. A day-care worker may have extensive and recent knowledge of a child’s behavior, whereas the mother can claim knowledge that is based on her social relationship. Indeed, once we begin to think of these things in this way, it is not difficult to see the [K+]/[K–] gradient as a simple heuristic that participants employ in order to operationalize what is in fact a much more complex set of normatively organized relations. This article has shown that a presumed, normatively accountable distribution of knowledge figures centrally in the formation and ascription of action, in the management of repair as well as in the negotiation of agreement in assessment sequences. Participants in interaction display a pervasive orientation to the distribution of knowledge across a wide range of different activities. Indeed, it would seem that epistemics constitutes a basic underlying order of relevancies, with consequences for the many other domains of interactional organization. SEE ALSO: Conversation Analysis, Overview; Ethnomethodology; Modality; Ques-

tions and Questioning; Turn-Taking

References Bolden, G. (2013). Unpacking “self”: Repair and epistemics in conversation. Social Psychology Quarterly, 76(4), 314–342. doi: 10.1177/0190272513498398 Clark, H. H. (1996). Using language. Cambridge, UK: Cambridge University Press. Durkheim, E., & Mauss, M. (1963). Primitive classification. Chicago, IL: University of Chicago Press. Foucault, M. (1980). Power/knowledge: Selected interviews and other writings, 1972–1977 (C. Gordon, Trans.). New York, NY: Random House.

EP I S T E M I C S

13

Heritage, J. (2002). Oh-prefaced responses to assessments: A method of modifying agreement/disagreement. In C. Ford, B. Fox, & S. Thompson (Eds.), The language of turn and sequence (pp. 196–224). Oxford, UK: Oxford University Press. Heritage, J. (2012a). Epistemics in action: Action formation and territories of knowledge. Research on Language and Social Interaction, 45, 1–29. doi: 10.1080/08351813.2012.646684 Heritage, J. (2012b). Epistemics in conversation. In J. Sidnell & T. Stivers (Eds.), The handbook of conversation analysis (pp. 370–394). Oxford, UK: Wiley Blackwell. Heritage, J., & Raymond, G. (2005). The terms of agreement: Indexing epistemic authority and subordination in assessment sequences. Social Psychology Quarterly, 68(1), 15–38. doi: 10.1177/019027250506800103 Marx, K., & Engels, F. (1845/1966). The German ideology. New York, NY: International Publishers. Pomerantz, A. (1980). Telling my side: “Limited access” as a “fishing” device. Sociological Inquiry, 50, 186–198. doi: 10.1111/j.1475-682X.1980.tb00020.x Raymond, G., & Heritage, J. (2006). The epistemics of social relationships: Owning grandchildren. Language in Society, 35, 677–705. doi: 10.1017/S0047404506060325 Robinson, J. D. (2013). Epistemics, action formation, and other-initiation of repair: The case of partial questioning repeats. In M. Hayashi, G. Raymond, & J. Sidnell (Eds.), Conversational repair and human understanding (pp. 261–292). Cambridge, UK: Cambridge University Press. Sacks, H. (1995). Lectures on conversation. Oxford, UK: Basil Blackwell. Sacks, H., & Schegloff, E. A. (1979). Two preferences in the organization of reference to persons and their interaction. In G. Psathas (Ed.), Everyday language: Studies in ethnomethodology (pp. 15–21). New York, NY: Irvington Publishers. Schegloff, E., Jefferson, G., & Sacks, H. (1977). The preference for self-correction in the organization of repair in conversation. Language, 53, 361–382. Retrieved from http://www.jstor.org/stable/413107 Schütz, A. (1946). The well-informed citizen: An essay on the social distribution of knowledge. Social Research, 13(4), 463–478. Retrieved from http://www.jstor.org/stable/40958880 Sidnell, J. (2012). “Who knows best?” Evidentiality and epistemic asymmetry in conversation. Pragmatics and Society, 3(2), 294–320. doi: 10.1075/ps.3.2.08sid Sidnell, J., & Barnes, R. (2013). Alternative, subsequent descriptions. In M. Hayashi, G. Raymond, & J. Sidnell (Eds.), Conversational repair and human understanding (pp. 322–342). Cambridge, UK: Cambridge University Press. Stivers, T. (2005). Modified repeats: One method for asserting primary rights from second position. Research on Language and Social Interaction, 38, 131–158. doi: 10.1207/s15327973rlsi3802_1 Terasaki, A. K. (1976/2004). Pre-announcement sequences in conversation. In G. Lerner (Ed.), Conversation analysis: Studies from the first generation (pp. 171–223). Amsterdam, Netherlands: John Benjamins.

Jack Sidnell is associate professor of anthropology at the University of Toronto, Canada. His research focuses on the structures of talk and interaction in ordinary and legal settings. In addition to research in the Caribbean and Vietnam, he has examined talk in court and among young children. He is the author of Conversation Analysis: An Introduction (2010), the editor of Conversation Analysis: Comparative Perspectives (2009), and coeditor of The Handbook of Conversation Analysis (2012; with Tanya Stivers), The Cambridge Handbook of Linguistic Anthropology (2014; with Nick Enfield and Paul Kockelman), and Conversational Repair and Human Understanding (2013; with Makoto Hayashi and Geoffrey Raymond).