Abstract
This article reviews arguments that, in the process of action formation and ascription, the relative status of the participants with respect to a projected action can adjust or trump the action stance conveyed by the linguistic form of the utterance. In general, congruency between status and stance is preferred, and linguistic form is a fairly reliable guide to action ascription. However incongruities between stance and status result in action ascriptions that are at variance with the action stance that is otherwise conveyed in the turn. This argument is presented, first, in relation to epistemic status and stance where the process is argued to be both fundamental and universal across all declarative and interrogative utterances. Some consequences of this way of viewing action are discussed. The argument is then briefly extended to deontics and benefactives.
Introduction
Students of language and discourse have long understood that grasping the meaning of utterances is profoundly influenced by contextual considerations. Some of this contextual influence is intra-sentential, and arises from the reflexive interactions of sentential constituents that constrain interpretation. Thus in a sentence like ‘He died old dying to be dead right’ (Ameka, 2008), a range of sentence constituents constrain the contextual reading of each successive use of ‘die’. Other aspects of contextual determination, of course, arise outside the sentence. Consider the well-known ambiguities of the sentence ‘The shooting of the hunters was awful’. One reading of the word ‘shooting’ might arise from an understanding that the speaker had just returned from a grouse hunt in Scotland, while another might be licensed by the knowledge that the speaker was concurrently reading the crime section of a newspaper. Similarly, ‘birds’ when described as ‘in the oven’ at Christmas or Thanksgiving are unlikely to be construed as owls, pigeons or robins (Rosch, 1975, 1977, 1978). In matters of the lexicon, as in other aspects of speaking, we achieve cognitive economies in the production of talk by relying on words and sentences to evoke the contextual specifications that recipients will use in understanding what we mean by what we say (Garfinkel, 1967; Grice, 1989; see also Heritage, 1984a).
In a celebrated body of work, Harold Garfinkel (1967) offered a sustained exploration of these issues by means of quasi-natural experiments. In one of them, the ‘conversation clarification’ experiment, subjects were invited to write down a simple fragment of their everyday conversation with partners or friends on the left-hand side of a page, and then to explicate the meanings that were conveyed in the fragment on the right-hand side. The protocol below represents the first two utterances of one student’s response to the task (Garfinkel, 1967: 38–39):
Garfinkel remarks that much of what is talked about here remains unspoken, putative common ground (Clark, 1996), evoked in and by the utterances and their sequential relations. And, in related ‘breaching’ experiments, he showed the anger and anomie that emerges when participants visibly fail to do the contextual infilling that is required to make sense of what is said, and thereby to make conversation work. Finally, in a systematic radicalization of Bar-Hillel’s (1954) observations about indexical expressions, Garfinkel argued that all language use was ‘indexical’ and ‘reflexive’. He meant by this that the semantic, pragmatic and actional interpretations of utterances are both unavoidably and irremediably shaped by their contextual background, and also transformative and renewing in their constitution of a current context of action and its future projections (Heritage, 1984a: 242). These notions were vigorously adopted by Sacks (1992; Garfinkel and Sacks, 1970) and incorporated into conversation analysis (CA).
CA’s primary operationalization of these ideas emerged in the examination of conversational sequences (Schegloff, 2007) and their relation to context (Schegloff, 1987, 1991, 1992a, 1992b). Analysis mainly focused on the role of overt, ‘on-record’ characteristics of earlier turns at talk in providing the necessary context for the elaborative understanding of subsequent turns. Indeed, in a justly famous discussion of ambiguity in the treatment of an utterance, Schegloff (1984) was able to show that two alternative readings of an ambiguous utterance that were taken up successively in a sequence were both potentiated by facets of the preceding turn and sequence. In this and other studies, conversation analysts were able to show that the meanings of current turns are shaped by actually demonstrable, rather than merely conjectured, aspects of context, and thus provided a powerful proof procedure for analyses that were lodged in sequentialized treatments of the context (Sacks et al., 1974). These analyses focused mainly, though not exclusively, on subsequent actions.
This treatment is nicely illustrated in the following well-known example from the CA literature. Here a relatively young child – Russ – has trouble in determining whether the question ‘Do you know who’s going to that meeting?’ (line 11) is a request for information from his mother, or a ground-clearing pre-sequence for his mother to tell him who’s going (Terasaki, 2004):
It can be observed that at line 7 Russ remarks ‘I know where yer goin’’, an utterance that, with a cooperative go-ahead at line 8, the mother treats straightforwardly as prefatory to Russ’s display of his informed status about the activities of his parents. It is this first sequence that, as Schegloff (1988) argues, primes and indeed licenses Russ’s understanding of the mother’s question at line 11 as prefatory to a similar and reciprocal display of her own. Once it is apparent that this is a misapprehension (line 13), it turns out that Russ has a ready answer (lines 15–16) to what, as it turned out, was intended as a ‘real’, that is, information seeking, question from his mother at line 11. This misapprehension, of course, is epistemic in character but it is grounded in, and indeed triggered by, the preceding sequence.
Other ambiguities and misapprehensions, however, may be grounded in contextual features of the interaction that are less clearly exhibited at the conversational surface. These may include the identities and other social characteristics that participants attribute to one another. In 1981, Susan Ervin-Tripp invited readers to consider a situation in which you are cutting up carrots with a large kitchen knife and a small child says ‘Can I help?’. And she continues: If you consider yourself the beneficiary of the assistance of a well-trained Montessori-taught carrot slicer, you may hear this as an offer. If you doubt the skill or even safety of the help, you may consider it a plea for permission, especially if said with a wheedling tone. The difference here is that in the case of permission requests the speaker, as principal beneficiary, wants the action more than the hearer . . . (I)n this situation there is no difference in the action itself which is in view. (Ervin-Tripp, 1981: 196–197)
In a similar way, the statement that ‘Someone just vandalized my car’ will likely be understood differently depending on whether it is part of a phone call to the emergency services (Wilson, 1991), or to other interlocutors. For example, as part of a call to a doctor’s office on the morning of an appointment, it may be heard as on its way to a request for a cancellation, as part of a call to a friend as soliciting sympathy, and so on, as in extract 3 (Heritage and Clayman, 2010: 65):
(3) ‘Someone just vandalized my car’ (Heritage and Clayman, 2010: 65; Wilson, 1991)
The point here is that the same utterance will attract different attributions of underlying motive or intention depending on the context of its production, the social roles of the participants, and the characteristics they attribute to one another (see also Levinson, 1981). These attributions will shape subsequent responses and the sequences through which they are implemented. However, their contextual grounds, though present to the speakers a priori, may not always become publicly available – incarnate in the talk – prior to the utterances whose interpretation they inform, where they would stand as conveniently overt and irrefutable evidence for conversation analytic observations and findings. Rather, they may only emerge a posteriori in sequences in which the relevant treatment of a prior turn is ratified as appropriate and anticipated. Another way to describe this situation is to invoke Schegloff’s injunction that contexts ‘external’ to talk must, if they are to be analytically defensible, manifest themselves within the talk – his ‘paradox of proximateness’ (Schegloff, 1991, 1992a, 1992b) – but to note that such contexts are often extruded out of the talk, ex post facto.
These considerations are apparent in circumstances where persons rely upon known-in-common attributes of one another as a means of constituting an utterance as a social action of a particular type. In the following sequence, for example, a doctor relies upon the patient’s authoritative knowledge of her marital status, and her rights to this authority, to achieve a request for information with a declarative utterance at line 5:
Here, as first observed by Labov and Fanshel (1977: 100), putatively shared knowledge of a recipient’s superior epistemic access to some state of affairs permits a speaker to produce, and be heard to produce, a request for information with declarative syntax.
In the same way, a ‘hearsay’ report about a recipient’s future plans is sufficient to evoke an informative confirmation:
That the turn in lines 1–2 was designed to request information, and that the turn in lines 3–4 was indeed informative, is confirmed by the questioner’s acknowledgment of the confirmation with a change of state token ‘Oh’ (line 4), registering a shift from lesser (hearsay) to greater (first-hand) knowledge (the ‘K− to K+’ shift; Heritage, 1984b).
In this article, I first consider processes of action formation and ascription (Levinson, 2012a; Schegloff, 2007) and review arguments for the role of epistemics in these processes. Subsequently, I will extend the ideas into other potential domains. Finally, I will describe some payoffs from this way of thinking about action formation, and set these arguments into a somewhat larger intellectual context.
Epistemics and action formation
Conversation analytic research on epistemics focuses on the knowledge claims that interactants register, assert, and defend in and through turns at talk and sequences of interaction. Research in this domain emerges from the confluence to two distinct streams of research. First, studies showed that interactants are strongly oriented to the prospect (Terasaki, 2004 [1976]), actuality (Goodwin, 1979) and retrospect (Heritage, 1984b) of information transmission in interaction, and that dedicated resources exist to address whether information that is possibly unknown to the recipient is in fact unknown, and ex post facto whether information that has been provided was in fact new, informative, and relevant for its recipient(s).
In subsequent developments, it was shown that participants are careful to index their relative epistemic positions vis-a-vis some state of affairs in sequences of interaction (Hayano, 2011, 2013; Heritage, 2011, 2012c; Heritage and Raymond, 2005, 2012; Raymond, 2003; Raymond and Heritage, 2006; Schegloff, 1996; Stivers, 2005). The position of an utterance within a sequence is particularly critical in these maneuverings. Participants who are engaged in responsive actions (in second position), but who may be more knowledgeable or authoritative (K+) about the matter at hand, may deploy a variety of resources to communicate this. Correspondingly, less knowledgeable (K–) parties, who find themselves making assertions about some matter that the recipient is more knowledgeable about, may implement a variety of practices to downplay claims to epistemic authority or primacy that might otherwise be tacitly communicated in and through these first actions.
If the first stream of research described above focused on the moment-to-moment management of relative claims to knowledge, the second addressed the epistemic background of utterances and, in particular, the impact of this background on how utterances are understood. Labov and Fanshel (1977) developed a five-fold classification of knowledge in relation to the local state of interaction: (6) [Labov and Fanshel, 1977: 100] A-events: Known to A, but not to B. B-events: Known to B, but not to A. AB-events: Known to both A and B. O-events: Known to everyone present. D-events: Known to be disputable.
This classification was developed to show that statements about B-events (e.g. ‘And you never called the police.’, p. 101) are understood as requests for confirmation even without the use of interrogative syntax or rising intonation (so-called ‘question prosody’). This argument found an echo in Pomerantz’s (1980) subsequent research on utterances that convey information that is derivative or hearsay – a basic distinction in languages that grammaticalize evidentiality (Aikhenvald, 2004). Pomerantz observed that when speakers frame knowledge claims in this way (as ‘Type 2’ knowledge) to a person with direct, first-hand knowledge of the matter at hand (‘Type 1’, knowledge), the utterance will be treated as a request for information):
Here Emma’s report of an event to which she has limited access (line 5) to a recipient who has more complete access constitutes an ‘off-record’ request to learn more (see Heritage, 2012b, for some extensions of this argument).
In a further extension of these ideas, Akio Kamio (1997) developed the concept of territories of information. Kamio mainly studied Japanese, a language that involves the systematic use of particles to index interactants’ relative ‘knowing’ and ‘unknowing’ status in relation to particular information domains (Kamio, 1997; see also Hayano, 2011, 2013). Kamio envisaged interactants as addressing domains of information that were ‘close’ or ‘distant’ to them relative to specific other co-interactants. As he conceived it, for each speaker an information domain could be arrayed on a continuum from 0 (highly distant) to 1 (extremely close and ‘possessed’ by the speaker). In addition to this thoroughly relativistic conception of access to knowledge domains, Kamio also broadened the concept of territory of information to embrace not only who knows what and in what way, but also who has rights to know it and express it, thus introducing a more fully sociological conception of knowledge domains (Abbott, 1988; Berger and Luckmann, 1967; Lukes, 1974; Pollner, 1974, 1975).
Epistemic stance and epistemic status: Action formation in polar requests for information
Action formation involves ‘the practices of talk and other conduct . . . which have as an outcome the production of a recognizable action X; that is, can be shown to have been recognized by co-participants as that action by virtue of the practices that produced it’ (Schegloff, 2007: 7). In the process of action formation, nothing is more fundamental than determining whether an utterance is delivering information or requesting it. Hinging on this determination is a whole range of secondary implications about whether an utterance is delivering news or renewing the relevance of given information, and the whole panoply of so-called ‘indirect’ actions: offers, requests, invitations, proposals and suggestions, and so on. In almost all the languages of the world, major grammatical distinctions are used to encode this fundamental distinction (Sadock and Zwicky, 1985).
However, in the specific case of polar requests for information, the relationship between grammar and action is not straightforward. First, about 18% of the world’s languages (including modern Greek and Italian; Rossano, 2010) have no morphosyntactic markings that single out interrogatives from declaratives (Dryer, 2008). Many linguists argue that in these languages questions are marked intonationally. But, as Levinson (2012b: 13) comments, ‘Every corpus study ever done on such languages, or on languages like English that uses declaratives as polar questions most of the time, has falsified this’ (see also Couper-Kuhlen, 2012; Geluykens, 1988). Moreover, even among many of the languages that have interrogative morphosyntactic markers for polar questions, a majority of questions are performed with polar declaratives (see Stivers, 2010; Stivers et al., 2010; Weidner, 2012, for a range of examples). Thus there remains a fundamental question of how declarative and interrogatively formed (polar) utterances are determined to be delivering or requesting information (Schegloff, 1984). In his discussion, Levinson (2012b: 13) observed that ‘the interpretative procedures are actually likely to be pragmatic’. In what follows, we pursue this line of reasoning, beginning with the distinction between epistemic stance and epistemic status.
The notion of epistemic status draws together the ideas of Labov, Pomerantz and Kamio. It begins from the notion that relative epistemic access to a domain will be stratified between two speakers 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 (Heritage, 2010, 2012a, 2012c; Heritage and Raymond, 2012). 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. While it may seem that it would be difficult for interactants to evaluate epistemic status relative to one another in domain after domain of information, in fact, as Labov and Fanshel (1977: 100) explicitly recognized, the thoughts, feelings, experiences, hopes and expectations of persons are generally treated as theirs to know and describe (Heritage, 2011; Sacks, 1984). 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 demonstrate this (Heritage and Raymond, 2005; Raymond and Heritage, 2006).
Epistemic status also, as Kamio (1997) stressed, involves not just the actual possession of information, but rights to possess it and to articulate it (Pollner, 1987; Raymond and Heritage, 2006; Stivers et al., 2011). It embraces what is known, how it is known (through what method, with what degree of definiteness, certainty, recency, etc.) and persons’ rights, responsibilities and obligations to know it (Drew, 1991; Heritage and Lindström, 1998; Maynard, 2003; Pomerantz, 1980; Stivers et al., 2011; Terasaki, 2004). For many domains of knowledge, the epistemic status of the interactants is an easily accessed, unquestionably presupposed, established, real and enduring state of affairs.
If epistemic status is conceived as a somewhat enduring feature of social relationships vis-a-vis an epistemic domain, epistemic stance by contrast concerns the moment-by-moment expression of these relationships as managed through the design of turns at talk. In English, epistemic stance is prominently expressed through different grammatical realizations of propositional content, evidentials and epistemic modality (Chafe and Nichols, 1986; Egan and Weatherson, 2011; Lyons, 1977; Palmer, 2001). Consider the following three utterances:
Each of (8)–(10) has the same propositional content (about the recipient’s marital status), but the epistemic stance encoded in the grammar of these three sentences is different. We can represent this in terms of the different epistemic K–/K+ gradients the sentences conventionally invoke (Heritage, 2010, 2012c). Each of these sentences addresses information that is well within the epistemic domain of the recipient. Thus the recipient has primary rights to know the information. However, (8) proposes that the questioner has no definite knowledge of the recipient’s marital status, and indexes a steeply sloping epistemic gradient between an unknowing (K–) questioner and a knowing (K+) recipient. (9) and (10), by contrast, express increasing commitment to the likelihood that the recipient is married, which can be represented by increasingly shallow ‘K– to K+’ epistemic gradients. (More elaborate operationalizations of a cline that extends into evidentials and epistemic modality as canvassed by Bolinger (1957) can be found in Levinson (2012b).)
In fact, the declarative question formats are mainly used when the speaker has already been told (or independently knows) the information and merely seeks to confirm or reconfirm it, or alternatively to convey inferences, assumptions or other kinds of ‘best guesses’ (Brown-Schmidt et al., 2008; Raymond, 2010; Stivers, 2010). More generally, in those languages that possess interrogative morphosyntax but also use declaratives to solicit information, the use of interrogative morphosyntax is concentrated in utterances that request new information, rather than the confirmation of existing information (Levinson, 2012b; Stivers, 2010; Stivers et al., 2010; Weidner, 2012). Further, while taking an ‘unknowing’ epistemic stance as in (8) invites elaboration and projects the possibility of sequence expansion, the more ‘knowing’ formats represented by (9) and (10) tend to invite confirmation and sequence closure (Heritage, 2002c, 2010; Heritage and Raymond, 2012; Heritage and Sorjonen, 1994; Raymond, 2010).
In general, speakers act so as to preserve, on the one hand, consistency between the epistemic stance they encode in a turn at talk and the epistemic status they occupy relative to the topic, and on the other hand, congruence between the expression of that status and the epistemic status of a recipient. Normally, both consistency with self and congruency with other(s) are realized when relatively unknowing (K–) speakers ask questions, and relatively knowing (K+) speakers make assertions. However, a variety of motives and contingencies may result in non-congruent utterances. Epistemic status can be dissembled by speakers who wish to appear more, or less, knowledgeable than they really are. And, as Raymond and Heritage (2006) demonstrate, interactional exigencies may compel, or simply eventuate in, divergences between epistemic status and epistemic stance (see also Hayano, 2011, 2013; Heritage, 2012c).
The idea that shared knowledge or ‘common ground’ (Clark, 1996; Enfield, 2006; Garfinkel, 1967; Stalnaker, 1978; Tomasello, 2008) concerning epistemic status is fundamentally involved in the production and ascription of action was first canvassed in Dwight Bolinger’s (1957) discussion of negative interrogatives. Considering a situation in which A is planning to hang out the washing in the yard, Bolinger observes that B’s remark ‘Isn’t it raining’ would be understood as a question if B could not see outside (a ‘blinds down’ situation), but would be understood as an assertion that objected to the plan if B could see outside (a ‘blinds up’ situation). This line of argument was subsequently pursued for negative interrogatives in a variety of CA studies (Heinemann, 2006; Heritage, 2002a; Koshik, 2002, 2005).
In a more recent systematic study, Heritage (2012a) shows that epistemic status is the central pragmatic resource in determining whether an utterance will be understood as requesting or asserting information and that it dominates the significance of linguistic form in this regard. A shortened version of his argument is summarized below.
Declaratives
In contexts where an utterance formed with declarative syntax concerns information that is (primarily) within the speaker’s epistemic domain, the speaker will be understood to be delivering information. For example, in (11), Ida has called her friend Jenny to tell her that some goods have arrived from the local department store, in a K+ declarative:
Here the declarative form of Ida’s turn is congruent with its content, which is exclusively within her epistemic domain. The result is an action that is unambiguously an ‘informing’ and that is acknowledged as such with ‘Oh’ (line 9) by her recipient (Heritage, 1984b).
By contrast, declaratives that address matters that are within the recipient’s epistemic domain are ordinarily construed as ‘declarative questions’ that invite confirmation (Labov and Fanshel, 1977; Quirk et al., 1975). For example, in the following sequence involving a discussion of Nancy’s ex-husband’s financial affairs, Emma offers a declaratively framed inference (marked with the inference marker ‘then’) about Nancy’s knowledge of her ex-husband’s whereabouts (line 11):
15 Nan: At a:ll, so: I jist assoom
Emma’s intervention is treated by both parties as a ‘question’: Nancy responds with a complex confirmation (Stivers and Hayashi, 2010), while Emma acknowledges this response (line 14) with a change of state (K– to K+) oh-receipt that acknowledges the information conveyed, and intimates her revised understanding of the situation (Heritage, 1984b). While inference markers clearly contribute to increased clarity about which speaker’s epistemic domain a particular item of information falls into, they are surely not essential to it, as a return to (4) indicates:
From time to time, errors premised on assumptions about epistemic domain can emerge. In the following case, A has recently returned from a fitness class that A and B attend on separate days.
B misinterprets A’s line 1 as a question, in terms of his presumed K+ access to his own fitness routine, and answers with a correction at line 2. At this point, A launches a third position repair (Schegloff, 1992c) that reveals that her utterance at line 1 was intended to deliver new information, and B in turn is duly corrected, acknowledging that correction with ‘oh’.
Finally, in the following, more complex, case CNN anchor Wolf Blitzer is managing a panel discussion of the health care reform act of 2010. This is a live discussion during voting in the House of Representatives, and a live feed on the screen indicates that the Democrats are presently three votes short of a majority. Blitzer’s turn at line 3, designed to establish agreement that the health care reform will be passed by the House, is deployed as a means of asking his addressed recipient – a Republican health care strategist – to comment on the next steps in the conflict.
However, shortly after Blitzer initiates his subsequent question (line 7) his recipient displays uncertainty about whether and how he should respond (‘Oh are you asking me or telling me.’), formulating this as confusion over whether Blitzer’s remark is conveying or requesting information. That the recipient might be confused is understandable. While Blitzer’s primary proposition (‘The Democrats are going to win’) was taken for granted among the panelists by this point in the broadcast, its frame (‘you realize of course’) targets a domain – personal knowledge – that is the privileged epistemic territory of a subject-actor. In short, the frame of Blitzer’s remark is at epistemic odds with its propositional content, and the outcome is a brief, but genuine, moment of confusion for the recipient.
The argument, in sum, is that the determination of when a declarative utterance is delivering information and when it is requesting it is contingent on the recognized epistemic statuses of the parties in relation to the substantive content of the declarative.
Interrogatives
In contexts where an utterance formed with interrogative syntax concerns information that is (primarily) within the recipient’s epistemic domain, the speaker will be understood to be requesting information. In the following sequence, a polar request for information by Nancy motivates a corresponding response from her friend Hyla. Here the information (about Hyla’s new boyfriend) is clearly in Hyla’s epistemic domain, and the status of Nancy’s turn at line 1 as an information request is reinforced and placed explicitly on record by the use of interrogative syntax and reinforced with rising intonation:
That this turn was produced from a K– position, and was to be heard as a request for information, is retroactively confirmed by Nancy’s oh-receipt of Hyla’s responses (line 3), which, once again, indexes a ‘change of state’ from K– to K+ (Heritage, 1984b).
Just as declaratives cease to deliver information when their content falls primarily within the recipient’s epistemic domain, so interrogatives fail to request information when their content falls primarily within the speaker’s epistemic domain. Two main classes of these anomalous ‘questions’ have been identified in the literature: ‘exam’ or known answer questions (Searle, 1969), and queclaratives (Sadock, 1971, 1974): assertions packaged as questions, for example, ‘Did I tell you that writing a PhD was a cakewalk’. Although exam questions seek to evaluate students’ understandings, they do not request informational content per se. In the following case, as the teacher’s various remarks make abundantly clear (lines 9–15), there is only one ‘K+’ person in the class who will determine the correctness of the students’ responses to this exercise in poetry appreciation:
Of course, the classroom context will undoubtedly prime the students’ understanding that the teacher has K+ status in the general domains of the questions he asks. However, this priming will not withstand a changed epistemic status associated with a changed epistemic domain. Consider a situation in which, hearing a loud noise, the teacher asks ‘What’s that?’. Under these circumstances, the students will almost certainly treat his question as a ‘real’ one in search of information. Moreover and more generally, all third turns that affirm or deny the correctness of students’ contributions (Drew, 1981; Sinclair and Coulthard, 1975) retroactively reassert the K+ epistemic position that informed the question’s production in the first place (Heritage, 1984a), just as ‘oh’ receipts and their cognates acknowledge a previously K– position.
In queclaratives (Sadock, 1971, 1974), interrogative syntax is used to frame utterances whose content is primarily known to the speaker. Here the speaker’s putatively K+ position cancels the possibility that the utterance will be heard as requesting information, but will rather be heard as ‘rhetorical’. The example below is a case in point. Shelley has cancelled plans for an out-of-state trip to a football game, and Debbie has called to protest this decision. The core complaint is that Shelley abandons her ‘girlfriends’ in favor of ‘guys’. In the following segment, Shelley is trying to deflect this accusation by indicating that the boyfriend would have paid for her to go if he were going (lines 2–3):
At line 5 Shelley asks ‘So an’ ↑when other time have I ever done
The argument, once again, is that the determination of when an interrogative utterance is requesting information is contingent on the recognized epistemic statuses of the parties in relation to the substantive content of the interrogative. In both cases – interrogative and declarative – epistemic status has a controlling influence on how the utterance will be understood. Reviewing a variety of grammatical formats for turns, Heritage (2012a) concluded that epistemic status was critical in interpreting their status as social actions. These conclusions are summarized in below:
(18) [Epistemics and action formation (Heritage, 2012a)]
In sum, there are many motivations for speakers to remain aware of their epistemic status relative to others on any issue under discussion. These include all of the issues falling under the topic of recipient design (Clark, 1996; Enfield and Stivers, 2007; Sacks and Schegloff, 1979; Sacks et al., 1974; Schegloff, 1972), and epistemic vigilance more generally (Sperber et al., 2010). However, the role of epistemic status in the production and recognition of actions adds a further, and significant, order of relevance for this consideration. If no declarative or interrogatively framed turn at talk can be fully grasped as an action without taking epistemic status into account, the organization of social action itself is profoundly intertwined with epistemic considerations. Correspondingly, monitoring epistemic status in relation to each and every turn at talk is an unavoidable feature of the construction of talk as action.
This use of epistemics in action formation involves a basic principle that much of human language production seems to follow: keep the signal simple and let the recipient (and the context) do the work of particularizing it (Levinson, 1987, 2000, 2006). As noted earlier, this principle is most clearly seen in lexis. According to Rosch (1975: 177), the human visual system ‘can discriminate some 7,500,000 different colors, but the most color names reported in any language are 4000 English names of which only 8 are used very commonly’. Color names are deployed by speakers in the expectation that they will be particularized by recipients by reference to context (Heritage, 1984a) using ‘fast and frugal’ heuristics (Gigenrenzer, 2004; Levinson, 1995), and so are prototype categories like ‘birds’ and ‘tables’, etc. (Rosch, 1975, 1977, 1978). Grammatical form may also take on prototype characteristics (Levinson, 2012b) and serve as a moderately reliable indicator of gross speech act types such as requests for information and confirmation (Stivers, 2010; Stivers et al., 2010; Weidner, 2012), and requests, offers, suggestions and proposals (Couper-Kuhlen, forthcoming), though these too are subject to massive contextual elaboration.
Payoffs from this treatment of action formation: Reducing theoretical demands on the linguistic signal
In this section, I want to comment briefly on what seem to me to be clear benefits from considering action formation as a product of both epistemic stance and epistemic status. The most prominent benefit may arise from reducing theoretical demands on the linguistic signal. Let me begin with an illustration from my own previous work on the uses of oh-prefaced second assessments (Heritage, 2002b). Broadly stated, I was concerned with the idea that oh-prefacing a second assessment was a way of showing that the second assessment is not a ‘here-and-now’ agreement to a first, but rather is grounded in previous and independent experience of the referent entity. By this means, oh-prefaced agreements are protected from the appearance of ‘mere agreement’ and disagreements are also likewise enhanced. In both cases the second assessment is solidified. The following is a case in point. Here two Orange County women are talking approximately one week after the assassination of Robert Kennedy in Los Angeles in 1968. Here, the matter on which they agree is not something of which the speakers have a directly shared experience, but rather something that, by virtue of the public media, they have experienced separately but in common:
In this example, Emma’s assessment of ‘the week’ evidently invokes the assassination event and its aftermath in southern California. Nancy’s oh-prefaced response indicates that she has separate and independent access to the assassination and its aftermath which Emma alluded to, and moreover that she has separately and independently arrived at the same conclusion as Emma. The paper argued that there were three primary drivers of these oh-prefaced second assessments: i) A first assessment can index or embody a first speaker’s claim to what might be termed ‘epistemic authority’ about an issue relative to a second, or to ‘know better’ about it, or to have some priority in rights to evaluate it. . . . ii) Moreover, a first assessment establishes a context in which a second can be found to agree or disagree. In such a context, respondents may be vulnerable to the inference that their response is fabricated on the instant to achieve agreement or disagreement, and is thus a dependent or even a coerced action within a field of constraint that is established by the first. iii) These issues can be compounded by the context of the assessment. The priority relationship between a first and second assessment may be less significant in contexts where the parties are joint experiencers of a state of affairs…. However where a state of affairs is separately experienced or known by the parties, going first can have a greater impact in implicitly establishing superior access, expertise, authority and rights to assess the matter in question, if only because the relative access and expertise of the parties to the state of affairs may remain to be negotiated. Where, out of these and related considerations, a second speaker wishes to convey that they have
However, oh-prefaced second assessments were also part of a process in which the second speaker asserted not only epistemic independence, but also epistemic primacy, as in the case below. In this sequence, Jon and Lyn are talking to Jon’s mother Eve. After Jon’s announcement about going to the movie Midnight Cowboy, Lyn asks Eve if she has seen it. She replies that she did not and goes on to account for this by reference to a friend, ‘Rae’, who reportedly said that the film ‘depressed her terribly’ (lines 5–6):
Clearly Jon and Lyn have seen the movie and have direct, rather than hearsay, access to its content: a classic position of epistemic primacy. Which position is oh-prefacing indexing: epistemic independence, or epistemic primacy? The solution to this problem is not to be found in the signal – the practice of oh-prefacing an assessment, but rather in the combination of the signal and the epistemic context it reflexively evokes. In the first case (19), direct access to the referent is shared, but experienced separately; in the second (20), direct access to the referent is not shared, and the epistemic situation that the practice indexes is correspondingly differentiated.
In a second illustration, I want to focus on the role of final rising intonation as a candidate practice associated with actions that ‘request information’. Cases in which final rising intonation is associated with requests for information are, of course, not uncommon:
However, cases in which final rising intonation is associated with speaker continuation are likewise very common. Indeed, so-called ‘continuing’ intonation is a commonplace of social interaction, as in the following case. Here Katherine wants to be picked up from a station on Monday, only to encounter an obstacle to her plans (lines 5–6). Expanding on the problem at lines 10–11, Lesley (Katherine’s mother) describes the arrangements for meeting ‘Granny off the bus’:
This informing with its prosodic final rise is understood as incomplete by Katherine, who acknowledges it with ‘ehYeah’, whereupon her mother continues with an alternative arrangement (lines 13–14). Whatever is accomplished by this final rise prosodic contour, it is not ‘questioning’. The mother is informing her daughter about arrangements, and her prosody invites a ‘continuer’ through which Katherine can indicate that she is tracking her mother’s utterance, passing on an opportunity to initiate repair, and thereby showing understanding and inviting the mother’s continuation (Schegloff, 1982; Stivers, 2008).
Then again, final rise – this time on a word – is associated with a practice known as ‘try-marking’ in which the speaker is trying to solicit a display of recognition from the recipient:
And final rise on a word can involve simply trying to solicit the attention and/or recipiency of the addressee for a variety of interactional purposes and projects:
The epistemic contexts of these cases of ‘final rise’ intonation vary considerably. In (21) and (22) the speaker is in a K– position, in (23) a K+ situation, and in (24) and (25) there is an issue of intersubjectivity: it is unclear whether the recipient recognizes a particular name. In (26)–(27) there is, as it were, a null epistemic position, or at least a situation in which ‘epistemics’ per se are irrelevant, for what is happening is some form of ‘summoning’. What all these cases have in common, however, is what Stivers and Rossano (2010) describe as a response mobilizing property. If final rise as a practice has an underlying ‘semantics’, it must be surely be to mobilize response. In this capacity, it can contribute an urgency to whatever interactional project is ‘in play’, and what that project is will be grasped, at least in part, by reference to the epistemics that are also in play in the moment. The identification of final rising intonation with ‘questioning’ is an unnecessary burden on the signal itself: unnecessary because it involves the attribution of a misplaced concreteness to a signal that is elaborated by a multiplicity of contexts.
Widening the analysis: Other forms of interplay between statuses and stances
I have so far argued that the interplay between stance and status can be a productive way of consideration issues of action formation at one of its most basic layers – the determination of whether an utterance is conveying or requesting information. But are there other elements of status and stance that could prove valuable in understanding the process of action formation.
Deontics
In a recent article, Stevanovic and Peräkylä (2012) have suggested that the proposal and acceptance of courses of action may be designed by reference to deontic status – the right to determine future courses of action. These proposals embody what they term deontic stance – the linguistic expression of relative rights to influence future courses of action. Deontic status and stance are every bit as nuanced and complex as their epistemic cousins, and their analysis in interaction is in its infancy (Stevanovic, 2012a, 2012b, 2013; see also Curl and Drew, 2008). Stevanovic and Peräkylä propose that, as in epistemics, there is a preference for congruence between deontic stance and deontic status – indeed that the management of social hierarchy itself may, in part at least, turn on this preference. Here, by way of a gesture towards this notion, I want to focus on a few cases involving incongruity between deontic stance and status.
Interaction, of course, involves many moments of deontic incongruity. When told of some exciting or surprising news, informed parties respond with ‘Shut up!’ or ‘Get out of here!’, they are trading on deontic incongruity to convey surprise or interest. Deontic incongruity is likewise not an uncommon feature of the closings of telephone conversations. The following is perhaps a prototypical case:
Commenting on this case, Schegloff and Sacks observe that: While B’s initial utterance in this excerpt might be grammatically characterized as an imperative or a command, and C’s ‘Okay’ as a submission or accession to it, in no sense but a technical syntactic one would those be anything but whimsical characterizations. While B’s utterance has certain imperative aspects in its language form, those are not ones that count; his utterance is a closing initiation; and C’s utterance agrees not to a command to get dressed (nor would she be inconsistent if she failed to get dressed after the conversation), but to an invitation to close the conversation. (1973: 313)
It may be added that the deontic incongruity between the deontic stance (D+) encoded in B’s utterance and B’s underlying deontic status significantly contributes to its formation as a pre-closing, and also to C’s recognition and treatment of it as such.
Benefactives
In a variety of environments deontics are closely associated with benefactives. Both concern future actions, but while deontics concerns at whose initiative an action will be undertaken, benefactives concern the determination of the benefactors and beneficiaries (put simply, who benefits) from some projected future course of action (Clayman and Heritage, forthcoming; Couper-Kuhlen, forthcoming; Levinson, 2012a; Searle, 1969). Benefactives can be a site of complex issues of rights and responsibilities (Bergmann, 1993; Raymond and Zimmerman, 2007; Zimmerman and Wakin, 1995). Some of the complexities involved are apparent in the following case, in which Skip has called a work associate (Jim) to secure a ride to work. At line 5, Skip’s utterance (‘You
Having agreed to the request at line 6, Jim asks ‘What time wouldju like the c
In a final case, Lottie offers to take her sister Emma to the beauty parlor (lines 1–2). It is clear that, at this point, Lottie is casting herself as the benefactor and Emma as the beneficiary of the offer. Emma has no car and cannot drive, and Lottie is proposing (with ‘come dow:n get chu’, line 1) to remediate this in the interests of a joint expedition. After Emma’s rather robust rejection of the offer (lines 4–5), Lottie revises the offer twice at lines 7–8 and 11 (Davidson, 1984):
Following Emma’s rejection at line 13, the sisters have reached an impasse and after a long pause of nearly a second (Jefferson, 1989), Emma comes up with her own offer at line 15. What is noticeable about this offer is that the joint activity it puts forward is virtually identical to Lottie’s earlier one at line 1. However, and crucially, it reconfigures the relationships of benefactor and beneficiary. Whereas Lottie’s original offer at line 1 framed Emma as the beneficiary (‘Don’t you want me to . . .), Emma’s offer at line 15 (‘If you
Conclusion: Bringing society back in
In a string of papers in the early 1990s on the general topic of talk and social structure, Schegloff (1987, 1991, 1992a, 1992b) urged that external (social structural) context be construed as part of the project of utterances: meaning that utterances are built to evoke the external contexts that recipients must use to arrive at a grasp of what they mean. The present argument parallels Schegloff’s. It is argued that utterances may be built to evoke, and be interpreted by reference to, known-in common mental predicates, social orientations and social rights, whether epistemic, deontic, or benefactive. However, as in the entire domain of interaction, participants must produce actions and respond to them based on the attribution of these understandings to one another. Their actions, together with the understandings that inform them, will stand or fall with subsequent events in the sequence. Thus in the case of understanding, unlike action, a secure grasp of the actions performed and the environment of understandings that inform them shares the fate of Hegel’s owl which, it will be recollected, flew only at twilight (Hegel, 1991: 23). For participants and conversation analysts alike, secured understanding will normally be achieved and validated after the event, downstream of the actions that realized the context in the first place.
This discussion of the role of status and stance in the process of action formation and ascription has clear echoes of Searle’s (1969) discussion of propositional content rules and preparatory rules in the implementation of speech acts. Here, however, a dynamic, sequential and mutually elaborative relationship between the two is envisaged. Within this more dynamic conceptualization, offers and requests are reconfigured (as in (30)) as a product of manipulations of the deontic stances of the parties, or (as in (31)) of changing understandings of the benefactive statuses of the parties. A ‘promise’ to give someone corporal punishment will, as Searle argued, be understood as a threat by one who does not relish the experience. But if recipients of such a ‘threat’ respond with ‘I look forward to it’ – thereby taking a stance in which they constitute themselves as beneficiaries of the action – the underlying status of the event may once again undergo alteration to conform with the more basic meaning of the term ‘promise’. In this way, the Searleian categories may be understood to have a complex and dynamic relationship within the actions that make up sequences of interaction, and perhaps attract the renewed attention of scholars of interaction. The issues are complex and much remains to be done. For example, while epistemic status was argued to be a nearly omnirelevant background in action formation (Heritage, 2012a), the (probably) more restricted circumstances in which deontic status or benefactive status are made relevant in the interpretation of utterances remain to be determined. Moreover, at the present time, the interactional dimensions and organizations for all three are more or less absolutely unknown.
Not least among the payoffs from this way of considering action formation is that it may offer ways to introduce considerations of social role, status and identity into the heart of action formation and ascription. Proponents of membership categorization analysis (e.g. Stokoe, 2012) have long urged that aspects of social identity and social organization that are ‘outside’ the talk are reflexively incorporated into the analysis of what contributions to interaction are accomplishing in terms of meaning and action. Insofar as elements of epistemic, deontic and benefactive status are construed as involved in the treatment of utterances as social actions, then attributions concerning identity and social structure will more easily, naturally and explicitly fall within the scope of the conversation analytic treatment of social action.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.