Reference without anaphora: on agency through grammar

1.1 Reference and anaphora in discourse

The study of linguistic reference has overwhelmingly focused on anaphora resolution as the clearest domain in which to investigate what, at least at the inter-sentential or discourse level, has been the central issue: linguistic and processing constraints on co-reference (see, e.g., King and Lewis 2017; Mitkov 2003 for overviews; and Geach 1962 for a historical dimension). Implicit in such studies is an orientation to the Gricean Maxim of Quantity (Grice 1975): that, canonically, an anaphoric form is parasitically referential by means of a link through an antecedent.^[1] And, in the data of naturally occurring language use, we see ample evidence of such forms.

Take as a case-in-point the following extract (1) in which Annie is first referred to as such (line 1), and then subsequently as she (lines 3, 5):

Once Annie has been introduced into this spate of talk, and given that there are no competing potential antecedents that could create referential ambiguity, both speakers can freely use the anaphoric reference she in subsequent utterances, as seen in lines 3 and 5.

The same is true of other NP referents and their corresponding anaphoric forms, as in (2) where an initial referent the house (line 1) is referenced subsequently with the anaphoric it (line 5):

This is similarly the case for references to places (3), times (4), and various other ontological categories.

Such paradigmatic cases in conversational data evidence, and appear to underscore, standard assumptions regarding co-referentiality. With respect to interaction, Schegloff (1996b), in discussing reference to persons, makes an initial distinction between reference forms and the occasions on which they are produced – in other words, the positions they occupy in discourse:

We can note that there can be locally initial reference occasions and locally subsequent ones – the first time in a spate of talk that some person is referred to and subsequent occasions in that spate of talk in which that person is referred to. And, separately, we can note that there are locally initial reference forms and locally subsequent ones. Full noun phrases, for example, or names can be used as locally initial reference forms … pronouns are transparently designed for use as locally subsequent reference forms. (Schegloff 1996b: 450 [italics in original])

Such a perspective finds common ground with Hopper and Thompson’s (1984) observation that the prototypical function of the noun phrase is to introduce a referent into a spate of discourse, in order to track that referent. Accordingly, most of a language’s complex nominal morphology will be used for that prototypical function, and other functions of nouns and noun phrases will tend to show reduced morphological alternatives (on which, see also Raymond and Fox 2020).

Whatever the theoretical or analytic commitment in the vast span of literature on (discourse) anaphora – be it dynamic semantics and DRT (e.g., Kamp et al. 2011) or natural language processing more broadly (e.g., Kilicoglu and Demner-Fushman 2016; Lappin and Leass 1994; Mitkov and Hallett 2007; Poesio 2016; Wolf and Gibson 2005), government and binding theory in generativist traditions (e.g., Büring 2005; Carnie 2013), Gricean and neo-Gricean approaches (e.g., Levinson 2000), the cognitive assumptions of Relevance Theory (e.g., Sperber and Wilson 1998), discourse-functional linguistics (Kärkkäinen 2003), or Hallidayan text analysis (e.g., Sanders and Pander Maat 2006) – the occurrence of such forms in a particular sequence, with an anaphoric reference form occurring subsequent to an initial antecedent, is taken as a given. Of course, Lasnik (1976) made the influential observation that anaphoric forms do not necessarily require antecedents, the interactional exemplification of which is captured in the following anecdote by Schegloff, who observes that, on the day that President Kennedy was shot:

One could walk on the street or campus and observe others being approached – or be approached oneself – by apparently unacquainted persons who asked, ‘Is he still alive?’ What was striking was that virtually without fail the reference was understood; and with great regularity that reference had taken the form of a … subsequent reference in … initial position. It served at the time as a striking embodiment of community, for each speaker presumed, and presumed successfully, what was ‘on the mind’ of the other, or could readily be ‘activated’ there. The … subsequent reference term tapped that directly … in the convergence of their orientations lay ‘community’. (Schegloff 1996b: 451)

However, even such exceptions to the assumed correlation of position and form – here, a subsequent form in an initial position – preserve the focus on subsequent reference forms. As a consequence, there has been broad consensus on the focus of linguistic attention – anaphoric reference – if not its treatment.

1.2 Non-anaphoric reference

Despite this overwhelming consensus from a variety of theoretical and methodological perspectives, investigation of reference in naturally-occurring language use reveals that, in fact, there are frequently occasions when this form-position paradigm does not hold in actual practice.

Returning to extract (1), for instance, we find the following continuation in (5), at lines 9–13:

Here, in line 11, after a flurry of anaphoric shes from both participants (lines 3 and 5), we see that the initial form Annie is re-introduced in line 11, and in a context where no competing antecedent has intervened that might create potential referential ambiguity with regard to the anaphoric form.

Linguistic accounts of such uses have hitherto emerged primarily from experimental studies and have been grounded in considerations of processing limitations – the assumption being that, over several turns-at-talk, successful co-reference requires the re-introduction of a full initial NP (see, e.g., Kazanina et al. 2007). Evidence from naturally occurring language use, however, reveals that speakers can – and in fact routinely do – successfully recover anaphoric references over substantial stretches of time. In discussing such “return-pops”, as they term them, Reichman (1981) and Fox (1987) both cite examples in which anaphoric forms are returned to unproblematically after several minutes of intervening talk.

By the same token, it is evident that processing limitations cannot be the sole determining factor in anaphoric usage, as speakers often reproduce initial (i.e., non-anaphoric) forms in the very next turn, thereby effectively repeating that initial form. In the following case involving the serving of pre-dinner drinks, for instance, the NP reference is to the ice:

In this case, the Gricean assumption that the second, subsequent reference to the ice is achieved by an anaphoric form (e.g., here, it, as in the case of the house earlier in (2)), is clearly not sustained.

Ordinary language use in fact displays an abundance of such exemplars where anaphoric reference could – and perhaps in Gricean and psycholinguistic terms, strictly should – but in fact does not occur, and so where a second full NP reference form is, strictly speaking, informationally redundant.^[3] A central irony here is that these exemplars are frequent in the environment of mundane conversation, which is standardly held – and indeed observed – to be a context replete with various forms of ellipsis. How, then, is it possible to account for such uses, which appear to challenge semantic, pragmatic, and psycholinguistic accounts of anaphoric reference?

In this paper, we use data from naturally occurring conversation in both American and British English to examine this use of full noun phrases in contexts where current linguistic accounts suggest anaphoric forms might otherwise be expected^[4] – what Schegloff (1996b) has referred to as “initial forms” in “subsequent position”.^[5]

This phenomenon has been documented in part by both Fox (1987) and Schegloff (1996b), who focus exclusively on person reference. In her foundational study of anaphora in English, Fox (1987) lists a number of these occurrences, ultimately concluding that such non-anaphoric reference forms occur in such a “wide range of conditions” (e.g., disagreements, overt recognitionals, assessments, etc.) that it is not clear there is “any common principle at work in them” (62). Moreover, the “interactive work” of this particular anaphoric pattern was “beyond the scope” of Fox’s study (64). Similarly, Schegloff, presenting some additional cases of person reference, and citing Fox’s work, notes that: “it quickly became apparent that a cluster of such instances occurred in disagreement environments of some sort. I … cannot take up the matter further here” (1996b: 453).

In the present analysis, we seek to extend these observations beyond reference solely to persons, and expand and consolidate these insights by bringing contemporary research to bear on this phenomenon. In pursuit of a unified account of such usage, our central finding is that what initially appears to be a divergent set of cases are in fact united by a common principle: the interactional expression of agency. What is more, the discovery of this principle has its origins in what might seem initially to be a wholly unrelated domain of work: linguistic stance.

3.1 Sequence-responsive: epistemic stance in confirmation and disconfirmation

One primary context in which non-anaphoric forms can be frequently observed in subsequent position is confirmation and disconfirmation. In these cases, their use is associated with the speaker’s primary epistemic access to the state of affairs under discussion, most commonly in responses to polar questions seeking information or confirmation, which we target here. Under these circumstances, the questioner has already ceded epistemic primacy to the respondent by requesting information from them, but the non-anaphoric reference form reinforces the respondent’s epistemic authority over the information at hand.^[10]

In the telephone-call extract (8), Margy has been describing some activities that her husband, Larry, is currently involved in: public relations and event planning for the city of Los Angeles. Emma then comments that she saw one such event described in the paper (line 10), to which Margy responds that Larry himself wrote the article: Mhm well tha:t’s Larry’s story. (line 11). After a silence, Emma seeks confirmation that it was indeed Larry who wrote the story, a routine method of noting the newsworthiness of an announcement and displaying interest in it (Jefferson 1993; Thompson et al. 2015: 80–82; Heritage 1984; cf.; Raymond and Stivers 2016). Observe the form of Margy’s confirmatory answer in line 14.

In response to Emma’s yes/no question at line 13, which might have received a default, type-conforming yes response (Raymond 2003), Margy prioritizes a repetitional answer that includes the non-anaphoric Larry’s story, providing the interjection answer yeah only after this: That was Larry’s story yeah (line 14). Moreover, it is noteworthy that in addition to reusing the initial form Larry (cf. anaphoric his), Margy’s answer also reintroduces the noun itself (story) which was not included in Emma’s question in line 13, thereby providing a morphosyntactically expanded response vis-à-vis the question (Raymond 2017; Stivers 2005). As is frequent with repetition confirmations (Heritage and Raymond 2005), Margy immediately expands her turn to further specify what she means by this being Larry’s story – namely that he wrote it – in overlap with Emma’s appreciation of this news in line 15. Crucial for our argument here is that the content of Margy’s confirmation in line 14 is not altogether new information; rather, it is information which Margy herself had already put forth in line 11. She is thus being asked for information that she herself has already stated (on which, see Schegloff 1996a), and it is in this sequential context that she asserts her primary epistemic rights over this information by way of a non-anaphoric and expanded repetitional answer.

Consider an additional case of this sort in (9), taken from a corpus of calls in which a reviewing doctor (REV) contacts ENT specialists (ENT) over the phone to discuss proposed courses of surgical treatment (Heritage et al. 2001). In discussing a particular patient’s condition, REV issues what eventually becomes, after a few self-repairs, a declarative request for confirmation with the perf (.) healed. (line 14–15) (Heritage 2012b). As in the above case (8), ENT has already asserted this in line 8 and is thus being asked to confirm a state of affairs that he himself has already put forth. In response, he provides a non-anaphoric and repetitional answer (line 16).^[11]

In this particular case, given that the reviewer is gathering information in order to evaluate the installation of tympanostomy tubes for this patient, the specialist may interpret the reviewer’s request for confirmation in lines 14–15 as a move toward a no-problem diagnosis (Heritage and Stivers 1999), thereby potentially undermining the attending ENT’s ability to obtain approval for this surgical procedure. The non-acquiescent, agentive nature of his non-anaphoric and repetitional answer effectively resists such a move: First, contrastive stress on perf extracts this noun from REV’s question to establish that while the perf(oration), specifically, may no longer be an issue, there is indeed still an issue with something else (Bolinger 1961; Couper-Kuhlen 1984; Ogden 2006; Schegloff 1998). This then provides for an expanded turn that insists on the necessity of the procedure for this patient: The perf’s healed but he had a (.) totally flat immovable (.) tympanic membrane (and beyond the data reproduced here). The agency of this particular non-anaphoric repetitional answer thus not only reasserts the specialist’s primary epistemic access to this particular patient, his medical history, and the appropriate treatment, but also thereby effectively resists acquiescing to a claim that might jeopardize his ability to get the surgery approved.

It is not only in confirming the content put forth in questions that non-anaphoric reference forms occur; disconfirmations, too, can be constructed with full NP referents preserved in their design in lieu of anaphoric forms. In such cases, the fact that the questioning turn’s polarity is inapposite may provide an additional impetus for answerers to agentively index their rights over the information in the question, with non-anaphoric forms being a primary means to accomplish this.

In the following case, originally presented in Fox (1987: 62), several men are dismantling a store and the topic turns to some fish tanks that may be divided between them. The topic is initiated by Mike (line 1), who appears to want one of the tanks, with a question to Vic, who is the informal leader of the group and whom Mike treats as the putative “owner” of the tanks. In response, Vic enumerates the range of tanks by their volume (lines 2–3), whereupon another participant, Rob, asserts that the tanks are owned by another individual, Alex. Rob begins his intervention with a declaratively formed assertion – But those were uh::: Alex’s tanks. (line 6) – which incorporates a stance of definite knowledge and is framed as a counter to Vic’s previous claim. Confronted with an “open-class” (Drew 1997) repair initiation (Hah?) from Vic, Rob renews his claim with a less definite assertion incorporating a tag question: Those’re Alex’s tanks weren’t they? (line 9) (see Table 1). After a second open-class repair initiation (Pardon me?), Rob reframes his assertion with a negative interrogative (line 11), which cedes further epistemic ground to Vic (Bolinger 1957; Heritage 2012b: 18; see also; Heritage 2002a; Heritage and Raymond 2021).^[12] It is in this context that Vic’s response (line 12) incorporates the non-anaphoric recycling of Alex.

In a context where Rob has already ceded the epistemic high ground to Vic by shifting from the declarative to the interrogative, it may be asked why Vic should insist on asserting it in line 12. Here it may be noted that the sequence began with a challenge to Vic’s right to dispose of the tanks (line 6); Vic’s response at line 12, then, is offered as a definitive riposte to that challenge. Vic immediately expands this riposte with another clause that incorporates a second non-anaphoric reference to Alex, asserting that Alex is trying to obtain tanks from him. In both units of talk, the non-anaphoric reference form is accomplice to an agentive rejection of a rival claim, a rejection that is based in Vic’s assertion of primary epistemic authority over Alex’s tank ownership.^[13]

In a second disconfirmation case, two women – Vera and Jenny, seen above in (7) – are making arrangements to car-pool to the local shops, when Vera asks whether a third, Ann, will be accompanying them. Jenny disconfirms the implication of the question (Bolinger 1978; Heritage 2010) with an account, while incorporating a non-anaphoric reference to Ann:

Here the additional assertive force of the response that is mobilized by the repeat of Ann in a subsequent position appears to be motivated by the fact that, from Jenny’s point of view, this should be shared knowledge between them (“common ground”, Clark 1996). Indeed, once Vera’s question has revealed that this is not the case, Jenny goes on to invite Vera to remember (line 11) that Ann had described her plans to them both, something that Vera does subsequently claim to remember in line 13 (Heritage 1984).^[14]

It is clear, then, that when confirming or disconfirming the state of affairs that a prior speaker’s question has put forth, second speakers have a range of linguistic and grammatical resources with which to agentively assert their own epistemic primacy or independence over the content under discussion. In the same way as repetitional and oh-prefaced answers diverge from the default type-conforming answers of yes and no by asserting some level of independence from the prior turn (Heritage 1998; Heritage and Raymond 2012; see also Table 1), the reintroduction of a full NP referent in lieu of an anaphoric form can work to similar effect, attenuating the dependence that the non-anaphoric answer has on the utterance being confirmed or disconfirmed and thereby agentively asserting the answerer’s primary rights to the knowledge at hand. In this light, we can see how the cases mentioned earlier and observed by Fox (1987) and Schegloff (1996b) to occur in the environment of disagreement, may be motivated by the same principle – disagreement being a primary context for the assertion of agency.

3.2 Sequence-initiating: deontic stance in sequence/topic initiation

Given the association between full, non-anaphoric reference forms and initial position, a second environment in which full NPs are recurrently observed is both in topic-shifts – where clearly there is a move away from one topic to another – and in the initiation of new sequences, where, although the topic might stay constant, there is a move to a new action within that topic.

Conversational participants typically shift to new topics in a “stepwise” fashion (Geluykens 1993; Jefferson 1984; Maynard 1980; Sacks 1992), thereby effectively “shading” (Schegloff and Sacks 1973: 305) the transition from one to the next. That is, speakers design new topics to fit the ongoing talk such that the prior topic was never actually brought to an analyzable end before the next one was launched.^[15] There are occasions, however, when a speaker may endeavor to launch a new topic in a more overt or disjunctive manner (Button and Casey 1985; Clift 2001; Maynard 1980; Jefferson 1978, 1984), particularly in environments when a current topic has undergone demonstrable attrition (Jefferson 1993), or has been oriented to as finished (e.g., Drew and Holt 1998). In producing such disjunctive shifts, speakers work to “boundary off” a current turn from the prior talk, conveying that the current turn should not be understood as a continuation of, or as having been occasioned by, what came before, but rather as the launch of something new. It was Fox (1987: 69–72) and Schegloff (1996b) who first observed the association of initial forms with new sequences of action, specifically in the domain of references to persons. We are now in a position to show that the selection of these initial forms is motivated by an assertion of deontic authority, and that this more general account extends to non-person NPs as well. We argue that such a move in interaction is inherently agentive in that it effectively proposes a new agenda to be addressed by the participants in subsequent talk; indeed, non-anaphoric reference forms are recurrently deployed in such turns as a grammatical means through which to assert this deontic authority. Thus insofar as anaphoric reference contributes to cohesion across turns (Fox 1987: 20–38; Halliday and Hasan 1976), here we will argue that non-anaphoric reference can function to attenuate or disengage that cohesion through the expression of deontic agency.

Consider the case of referring to Annie, presented earlier and repeated here as (12). While Maude might have continued using she to ask the question in line 11, she instead elects to use the initial form Annie.

Prior to Maude’s question in line 11, the participants have been discussing the senses of humor of their friends. This topic begins to undergo attrition across lines 7–10: Maude receives delayed and only minimal uptake from Bea in line 8, and then again in response to the upgraded version of the claim in line 8, thereby clearing the way for a topic shift (Jefferson 1993). It is in this environment that Maude launches a new topic with the sequence-initiating question .hhh We:ll: anywa:y <Was Annie there last night (line 11). Various components of the design of this turn constitute its topic-shift-implicative nature, including the audible in-breath in pre-beginning position (Schegloff 1996c), the well-preface (Heritage 2015), as well as the shift-implicative anyway (Ferrara 1997; Halliday and Hasan 1976; Park 2010; Sacks 1992: 561–569; see also Drew and Holt 1998: 510). To this list of features, we can also add the use of the initial form Annie, repeated here in place of the anaphoric she. While Annie had already been introduced, quite recently in fact, into the conversation in the context of discussing senses of humor, at this point Maude launches a discussion of last night’s card game, and specifically Annie’s participation therein. In this case, this shift in topic also constitutes a shift in granularity (Schegloff 2000), moving from assessing Annie and her sense of humor as a whole or in vacuo, to discussing her participation in a particular event. By attenuating her turn’s dependence on prior talk through the use of an initial (rather than subsequent) reference form, Maude not only presents her question as not having been occasioned by what came prior, but in so doing, she asserts a deontically authoritative stance in that subsequent participant conduct must now be oriented to the new topic that she has put forth, as opposed to a continuation of the line that the participants were discussing before. Exercising such deontic rights in interaction is, we argue, an inherently agentive move, and here we see specific grammatical means through which that agency is sequentially accomplished.

Consider an additional case of this sort in (13), previously analyzed both by Fox (1984: 71) and Schegloff (1996b: 452). Here, Mark and Karen discuss that girl [Keith] use to go with (line 5), whom Karen identifies as Alice in line 6 (Sacks and Schegloff 1979). Alice is subsequently referred to anaphorically with she and her as Karen offers up what she knows about this individual’s current whereabouts (lines 10/12). Similar to the prior example, though, this topic (i.e., Alice’s whereabouts at present) begins to lapse across lines 13–15, as Mark provides only minimal uptake to Karen’s second version of I dunno where she is. It is in this environment that another speaker, Sheri, self-selects (Sacks et al. 1974) to offer an assessment of the individual in question, referring to her with the initial form Alice in lieu of the pronoun she (line 16).

While Alice was unambiguously the subject of the prior sequence, she was topicalized in terms of her relationship to Keith (line 5). At line 16 and beyond, Alice emerges as the subject of a novel action trajectory. This is promoted by Sheri’s initial assessment (line 16) with its repetition of Alice, the initial reference form serving to disengage the continuative link that an anaphoric reference form would otherwise have engendered. And thus rather than a continuation of the prior discussion, Sheri treats this as “a new unit … separate from what has gone before” (Fox 1987: 71), “embod[ying] this, and incipiently constitut[ing] it, by use of the … initial reference form” (Schegloff 1996b: 452). That Sheri self-selects to offer this perspective further underscores the agentive nature of such an interactional move. In this case, launch of this new action trajectory is successful in that Mark first agrees with Sheri’s assessment (very o:dd) and then immediately goes on to provide what is offered up as specific evidence of Alice’s “strangeness” (lines 18–20), produced with anaphoric reference forms.

As we have already seen, it is not only in reference to persons that initial, non-anaphoric forms can be used in subsequent position, and thus the present analysis expands the scope of prior work by accounting for subsequent NP references to other ontological categories as well. In the following example (14), Hyla confesses that she phoned a romantic interest of hers (Richard), but immediately hung up when he answered the phone (line 3). Nancy then interdicts Hyla’s cost-based account for having done so (lines 6–7, 10–11, 13, 15, 17) to ask how she knew that it was in fact Richard who answered the phone: How do you know he answered, could you tell his voice? (lines 14, 16). Likely due to this turn being produced in overlap, Hyla issues an ‘open-class’ repair initiator (Hu:h?) in line 19 (Drew 1997), thereby inviting Nancy to repeat her question, which she does in the immediately subsequent line 20. Hyla’s response in line 21 – Yeah, I knew his voice – then both affirms Nancy’s candidate understanding as well as provides an answer to the original question of How do you know he answered. Observe how his voice is then immediately reintroduced in the new sequence that follows this exchange, as Nancy proposes the emotional experience of hearing the voice as a topic for subsequent commentary by Hyla.

Following the question and its answer in lines 20–21, Nancy offers the change-of-state token oh which effectively closes this prior sequence (Heritage 1984; on such “sequence-closing thirds” more generally, see Schegloff 2007: 118–142). Following this, Nancy shifts the topic of conversation away from the factual events of the telling, to the subjective, emotional experience that those events engendered in Hyla.^[16] Despite his voice having just been mentioned in the prior sequence, Nancy reintroduces it in line 24 with an initial (i.e., full NP) form as accomplice to, and constitutive of, this shift in topical focus. The grammar of the turn thereby claims the deontic rights associated with launching a novel interactional trajectory, to which Hyla acquiesces with her answer in lines 25–26.

In naturally occurring language use, non-anaphoric reference forms are frequently found in such topic-shift and sequence-initiating environments – that is, after a referent has already been introduced and successfully understood, and in many cases after the referent has also been subsequently referred to anaphorically across various turns-at-talk. In such cases, following previous accounts by Fox (1987) and Schegloff (1996b) but extending beyond references solely to persons, the initial, full NP form works to disengage or attenuate the cohesion that a pronoun would embody, thereby effectively bracketing off the prior topic or sequence in which the referent was implicated, and reintroducing the same referent for a new conversational purpose. Given that such turns effectively work against cohesion and establish a novel interactional trajectory for the participants – in that subsequent talk will orient to, and be interpreted with respect to, this shift in topic and/or in action trajectory – we argue that use of non-anaphoric reference in such contexts embodies a local stance of deontic authority, and as such exemplifies the agentive nature of this particular linguistic practice.

4.1 Epistemic rights beyond responding to questions

Confirmation and disconfirmation can occur outside the specific context of requests for information/confirmation that we have discussed thus far; that is, utterances that are not designed to request information, but rather to assert information,^[17] can also be agentively confirmed or disconfirmed by second speakers, with initial forms routinely deployed in the service of claiming primary epistemic authority over the content under discussion.

Consider the following case (15) in which the participants – fourteen-year-old Virginia, her older brother Wesley, and their mother – have been discussing a boy named Paul Paget. Mom has made it clear that she does not like this individual, an opinion which she defends by citing conversations she has had with another parent who has a daughter named Donna: Donna’s mother apparently found a letter that Paul had written to Donna and was not pleased with its contents. Mom has overtly expressed that she does not want her own daughter (Virginia) to interact with Paul either (data not shown). It is in the context of discussing Paul’s alleged interactions with Donna – and negatively assessing Paul more generally – that Mom asserts that Donna doesn’t like him (line 1). Observe Virginia’s confirmatory treatment of this assertion in line 2.

Virginia launches her response to Mom’s assertion with I know, thereby explicitly asserting her own, independent epistemic access to the content of Mom’s utterance (Mikesell et al. 2018). She then goes on to repeat the entirety of Mom’s turn, including use of a non-anaphoric Donna. This is combined with a turn design that is morphosyntactically and prosodically expanded compared to Mom’s initial assertion, resources that have been shown to similarly claim epistemic rights in responsive position (Stivers 2005; see also Raymond 2017). Thus the entirety of the design of Virginia’s turn-at-talk works to agentively convey that she has not been informed by Mom’s assertion in line 1, but rather possessed full knowledge of Donna’s dislike of Paul prior to Mom having said so.

Thus far we have seen that when first speakers request information, or endeavor to assert information, second speakers can use non-anaphoric reference forms to grammatically reassert their own superior knowledge over the proposition under discussion. An additional action environment in which this same practice is routinely observed in conversational data is in assessment sequences (Raymond and Heritage 2006), as seen in the following examples (16) and (17).

In the first of these cases, a grandmother (Vera) of two boys (James and Paul) who recently visited her home is discussing their behavior during the visit. Vera’s interlocutor is her neighbor and close friend Jenny, who is quite well acquainted with the grandsons. At line 1, Jenny begins with a positive assessment of one of the boys using a right-dislocation, projecting the name of the boy sentence-finally. As she approaches the end of the sentence, Jenny exhibits difficulty in arriving at the relevant name, eventually settling for James (line 3). However, at line 4, and in overlap, Vera supplies the other child’s name (Paul), and Jenny immediately corrects herself, replacing the reference to James with a reference to Paul. Perhaps to remediate this momentary faux pas, Jenny goes on to venture that James’s a little devil (line 9), thereby drawing a sharp line of demarcation between the two children, and clearly demonstrating that she knows ‘which child is which’. This demonstration, however, involves Jenny taking the lead in asserting a derogatory assessment of the child, an action that implies greater rights to perform the evaluation (Raymond and Heritage 2006).

As it turns out, Vera is in broad agreement with Jenny’s assessment, and is even prepared to upgrade the evaluation from little devil to little bugger (line 12). At the same time, and in keeping with her epistemically privileged position as a grandmother in the evaluation of the children, she deploys a non-anaphoric reference form (James) as a part of this assertion. Paralleling what we have seen in prior cases, the non-anaphoric reference form works to frame Vera’s assessment as not being produced “merely ‘in agreement with’ or ‘in conformity with’ the first speaker’s opinion” (Heritage 2002b: 219), but rather as an opinion that Vera herself had already arrived at independently – that is, prior to Jenny’s first assessment in line 9. With this more agentive usage, Vera indexes her greater epistemic access and rights to evaluate the children, underscoring in this process her status as a grandmother (Raymond and Heritage 2006). Use of the tag question at the end of this second-position assessment further supports such an analysis, as now it is Jenny who is placed sequentially in a position to agree with Vera’s assessment of the grandchild, rather than the other way around; and indeed, Jenny does just this in lines 13–14.

In an additional assessment case (17), Mary expresses surprise that a friend is dating somebody from the group (line 1). Alan supports and upgrades this point of view by singling out Tony as an especially surprising choice (line 6), going on to defend and account for that opinion after a silence (Maynard 2013). This account is abandoned, though, as Mary intervenes to offer a mitigated but positive assessment of Tony, reissuing the initial form: I don’t know Tony’s got a nice personality, (lines 9–10). Then, in response to this assessment, we see Alan reissue the initial form Tony again in line 11.

In line 11, Alan’s second assessment reissues the non-anaphoric form Tony as part of a modified repeat of Mary’s assessment. In addition, the initial agreement portion of the turn is oh-prefaced, thereby indexing Alan’s independent arrival at this same assessment of Tony and his personality (Heritage 2002b). That is, Alan is not simply “agreeing just to agree” in the moment, but rather presents himself as having already come to this opinion of Tony before Mary’s having just said it. This constellation of agentive grammatical practices thus work to upgrade Alan’s epistemic stance, disengaging the expression of his opinion from being taken as occasioned or coerced by Mary’s positive assessment.^[18]

Consider one additional action environment in (18), this instance concerning reference to an object rather than to a person. In this case, pictures of nineteen-year-old Emily have surfaced on social media in which she can be seen wearing an “All Saints”-brand jumper that belongs to her mother, Jane. Here Jane first solicits Emily’s attention in line 1, and then requests – in declarative form – that she return the jumper with Rea:::lly like you: to give me: my All Saints jumper back (lines 3–4), a request that simultaneously brings off an accusation. Emily’s response then preserves the full NP All Saints jumper in the immediately next turn, in lieu of an anaphoric it.

Emily’s response at line 6 – I don’t have your All Saints jumper – resists endorsing the presupposition in Jane’s prior request that she is in possession of the jumper. In extracting and exposing this implicit presupposition, she mobilizes the agentive character of the full NP in order to launch a full-bore denial, asserting primary epistemic authority over what she herself has/does not have, and thereby refusing to acquiesce to the accusation. Just as (16) and (17) show speakers laying claim to prior, independently reached assessments, this case shows a speaker pulling epistemic and chronological rank by means of a full NP in subsequent position.

While prior work left open the question as to why a seemingly diverse range of actions would be realized with non-anaphoric reference forms, examples such as those analyzed in this section reveal the crucial import of epistemic stance to the effectuation of such actions. Whether responding to a question, an assertion, an assessment, or request, second speakers risk being taken as occupying an epistemically inferior position vis-à-vis the first speaker – as simply acquiescing to the state of affairs presented in the first speaker’s question or request, as having been duly informed by the first speaker’s assertion, or as simply going along with the first speaker’s assessment. Use of initial reference forms in response to such first actions indexes firstness, and as such, agentively attenuates the dependence of the responsive action on prior talk in the service of asserting epistemic primacy or independence over the content under discussion.

4.2 Deontic rights beyond sequence and topic initiation

While we exemplified the expression of deontic rights with sequence and topic initiation, underscoring the role that initial NPs play in claiming those rights, close examination of the data reveals that deontic stance can become relevant in a range of other action environments as well. In this section we target two exchanges with NPs that refer to objects in the physical world, and we focus on the deontic stances claimed in the context of ongoing embodied courses of action.

The first case (19) comes in response to a question. A family is laying the table in preparation for a meal at Mary’s house. Mary’s adult daughter, Vanessa, has cooked for them and a visiting friend (Chris) of her younger sister, Susan. At line 2, Vanessa launches a question projecting the availability of something which Susan, in her response at line 7, identifies as (place) mats. This is subsequently acknowledged as essentially correct by Vanessa – if not the exact completion she was about to produce herself (Lerner 1996) – in Vanessa’s extra mat, and then, after a micropause, by a reformulation of a request in its entirety at line 8: is there an ↑extra mat ↓somewhere,. After a silence, Mary responds, not only in the affirmative, but also with an appended account: I found you a mat in he:↓re. In doing so, and in going beyond a simple type-conforming yes or no response (Raymond 2003), Mary effectively stakes a claim to having attended to the problem raised in the question, and solved it, before Vanessa’s raising of the matter at line 8. It does so by means, initially, of the stress on found, which indexes a contrast with the assumption conveyed in the recipient’s question (Raymond 2017). The use of the full initial NP a mat, rather than a subsequent proform one, here works to push back against the secondness of the turn. This claim to deontic agency by Mary, as de facto host, is further underscored by Mary’s subsequent inquiry of her daughter as to the purpose of the extra mat, which effectively requires an account from Vanessa.

Here the entitled and authoritative deontic stance that Mary assumes in line 10 with respect to the task at hand – laying the table – is thus evidenced in a constellation of features, of which the agentive use of the full form in subsequent position is a significant element.

Consider one final exemplar in (20). Once again, the setting is Mary’s house. Here, Vanessa initiates an offer sequence to her mother, Mary, using the default forms in their default positions: What did you do with your glass Mum … I’ll fill it (lines 5/9), the subsequent anaphoric form it following on from the initial NP your glass. At line 12, Mary is directing Vanessa’s pouring of the drink: Very little thanks … Thanks. Some moments later, at line 21, Adam, Vanessa’s boyfriend, who is visiting the family for the first time, launches an offer, stopping short of naming what he is offering to get and subsequently instead producing an account for why, after all, he cannot deliver on it, in the course of which naming the object of his offer: Dunno where the ice is actually s(h)o-, the appended action prompt so (Raymond 2004) explicitly formulating this account as a recruitment of assistance (Kendrick and Drew 2016). In overlap with this prompt, Vanessa receipts the account with the change-of-state token oh (line 23) (Heritage 1984), suggesting that her attention has just been drawn to this omission, and then resolves the issue by producing an offer, formulated as a report: Oh I’ll get the ice. Both this action and its formulation are highly agentive in character. In her response to his account, Vanessa – who might, after all, have simply facilitated Adam in the task of getting ice by directing him to the freezer in another room – effectively supplants him, taking over the task herself, which she does even despite being in the midst of chewing a bite of food. This action, deontically agentive in itself, is underwritten by the repeat of the full NP, the ice, in subsequent position:

In both (19) and (20), then, the initial form in subsequent position (a mat/the ice) is one means by which speakers make relevant and consequential particular identities in terms of the rights and responsibilities that are associated with them. Indeed, we saw this previously in Example (17), in which Vera underscored her status as a grandmother through her claim to have primary rights to assess the children in question (Raymond and Heritage 2006). Similarly here, in cases (19) and (20), the agentive second-position responses given in each case display their producers’ concern to position themselves as not merely responsive to, but as actively having anticipated, the requests in prior turn. In (19), Vanessa has cooked the meal in her parents’ kitchen. Her mother Mary’s agentive response to Vanessa’s inquiry about an extra mat shows her pushing back against the secondness of her sequential position; her assertion that she has found … a mat clearly reports an event – an engagement with the duties of a host – that antedates Vanessa’s request for one in line 8. In (20), Vanessa’s response to Adam at line 23 actively takes over the task he, the guest, has proposed – a quest for the ice. Adam’s incipient offer at line 21 potentially intercepts a line of action that Vanessa has initiated – the offering (line 9) and pouring (line 12) of Mary’s drink. Adam’s incipient offer introduces the possibility of ice, thereby potentially exposing Vanessa’s inattention to having, thus far, provided some. Vanessa’s report that she will get the ice, with its full NP in subsequent position, thus resists the initiative that Adam’s turn has displayed, and in so doing resists the responsiveness of its position. Further evidence for this analysis can be found in the oh-prefacing of Vanessa’s turn, which in this context treats the prior question as inapposite (Heritage 1998).

In (19), it is the host who designs a responsive turn to exert agency from second position. Example (20) might initially look to be a similar case: Vanessa’s orientation to the fact that she is Adam’s host, in that they are staying at her parents’ house. However, it is clear that in this case, the speaker is resisting the incipient offer of a guest who is implicating himself in the serving of the drink to his host, Mary. A local contingency is here relevant: Vanessa has, at lines 5 and 9, initiated the serving of the drink but has yet to display attention to the provision of ice until her response to Adam’s incipient offer at line 21. Her Oh I’ll get the ice, resisting Adam’s initiative to effectively take over the task he has volunteered to do, is thus hearable as following through, and seeing to its conclusion, a course of action that she herself has initiated—what Kendrick (2017) has called a preference for self-over other-remediation. The local sequential contingencies which shape the participants’ actions here thus seem less an orientation to the duties of a host as opposed to a guest, than wresting back the initiative with respect to her own course of action. In both (19) and (20), then, it is evident that speakers who are in a sequentially subsequent position – that is, responding to the initiative of others (finding a mat, getting some ice) – use full initial NPs as a means of exerting deontic authority over the trajectory of the interaction from those responsive, subsequent sequential positions.