I'm Sorry “About That”: Apologies, Indexicals, and (Unnamed) Offenses

We investigate an apology format, “I'm sorry about it/that,” where indexical terms (pronouns) refer to the offense rather than naming it. We identified two subsets in our collection of indexical apologies. In one, indexicals are subsequent either to the offense formulation or to an apology-relevant event; in the second, indexicals are used without any prior named offense. We found these two uses are associated to distinct contextual features. In the first set, apologizers also initiate the course of action, making relevant the apology; in the second, would-be apologizers find themselves at fault during/because of the call and apologize in response to a course of action initiated by the virtual offended person. Here indexicals draw their meaning from prior talk, serving speakers' interactional needs, when apologizing is problematic. More specifically, speakers manage to apologize without specifying the offense, often rejecting their responsibility for the virtual or actual offense.


INTRODUCTION
It is well-known that on some occasions simply saying sorry in not enough; apologizing properly may not just be a matter of expressing regret through a single word. Offering an apology involves selecting the form and the moment most appropriate to the circumstances in which the apology is due, chosen from among a variety of formats, some brief and some more extended, with accounts or without, mitigated or intensified (Biassoni et al., in press; Heritage & Raymond, in press). It may also matter how an apology is introduced in the course of a telephone call, for instance, whether an apology is a reason for the call or interactionally generated during the call (Drew, 1984;Sacks, 1992). In the first case, people apologize as the upshot of their own prior judgment but in the second as a consequence of some problematic circumstances arising from or evoked during the call. Another important dimension, and only partially overlapping the former, concerns the issue of who initiates the course of action leading to the apology, whether this is the apologizer or the virtual offended person. These two distinct situations entail the issue of whether, owing to its deployment in the conversation, the apology can be understood more as the result of an individual resolution (the apologizer's) or as the final upshot of a course of action initiated by the virtual offended person and coconstructed by the two participants. 1 These dimensions are relevant for the way in which apologies are shaped and constructed, including whether or not speakers describe or name the offense or whether instead they refer to the offense through the use of indexicals (i.e., "indexical expression/s" or "indexical term/s") (see also Drew & Hepburn, in press). This is the particular focus of the investigation in this article: the format of apologies designed as I'm sorry about X, where X is a deictic term, pronoun, or a time reference. We called these apologies "indexical apologies" (IAs).
So, in terms of their format, IAs differ from the other apologies by being constructed (1) through more than the single word or phrase sorry/I'm sorry but, (2) although an offense is referred to, that reference is unspecific and does not name the offense. This provides a very limited lexical inventory for characterizing the offense. In our collection, it includes the following options: the third person neutral pronoun "it," the demonstrative pronoun "that," and two time references "yesterday"/"tonight." In what follows we compare the difference between naming the offense (Cirillo et al., in press) and referring to an antecedent formulation of the offense. In the first type the apology is designed as one turn or part of it; in the second, the actual apology refers to some prior anchor point in the preceding talk.

NAMING VERSUS REFERRING TO THE OFFENSE
In example (1), Leslie names the offense by describing it as having failed to phone Philip the day before (line 8).
By contrast, in (2), with the expression "about that" (line 13), Emma refers back to "keep] people's pa':r too:::ls" (line 10), establishing a direct reference to a prior full-form referent in a context where the repetition of the formulation would indeed seem redundant and unnecessary.
However, in our collection, IAs are not always shaped by the preceding sequence, as above. The format in which the apologizer provides a description of the state of affairs that retrospectively becomes an offense (full named offense) and then refers to it with an indexical is present only in (2). In most other cases of our collection, as in (3) below, indexicals are used without any prior full formulation of an event or conduct that would work as explicitly "naming the offense." Leslie calls Joyce to enquire about whether she will attend the meeting that evening. Joyce answers negatively (line 4) and, later, produces an IA (line 8).
Although it is unquestionable that Joyce apologizes for not going to the meeting, in the interaction there isn't any explicit formulation of the offense. Indeed, Joyce mentions that she won't go to the meeting (" · hhh No I'm not Leslie.", line 4), and the indexical "that" in her subsequent turn in line 8 can be heard as referring to her backing out. However, if we compare the way in which the anchor point of the indexical (the virtual offense) is produced in (3) and in (2), we realize that although in (2) the offense ('keeping other people's power tools') is fully formulated and offered as Emma's own resolution, in (3) the failure to attend the meeting is offered in response to Leslie's enquiry (elicited information-giving).
We argue that when a full antecedent formulation is missing, the indexical acquires part of its meaning from all that is said before, due to the semantic emptiness of the indexical itself; thus, the sense of the offense is derived from where the apology occurs in the course of the conversation. Rather than referring to a name, a noun phrase or a sentence, the indexical reaches back to the whole preceding talk or to parts of it. In this way, the limited linguistic resources provided by indexicals to characterize the offense are balanced by their mobility, so as to incorporate unlimited adjunctive meanings, drawn from prior talk.
This mechanism is clearly enacted in (3), where the meanings of Leslie's request (line 2) and of Joyce's negative response (line 4) coalesce in the IA. It is, in fact, because Joyce apologizes that her prior negative answer comes to be understood not as mere information but as an offense. In turn, because of the apology, Leslie's question is not treated anymore as a request for information about Joyce's whereabouts (as in line 4), but it acquires a strategic nature, turning it into an attempt to enlist her in a prearranged meeting or perhaps as a presequence to some other projected actions. 2 We found that apologizers methodically use this type of reference, that is, deictics/indexicals without a prior named offense, in situations where the apology is interactionally generated and produced as a result of a course of action initiated by the recipient of the apology, the one who is "virtual offended." We argue that the features of indexicals described above-their basic semantic emptiness as well as their being amenable for semantic enrichment drawn from prior talk-accommodate the would-be apologizers' needs to finetune their response to their recipients, thereby tailoring the range of the offense to the circumstances of the interaction. Depending on where in the interaction the IA is produced, by not naming the offense speakers manage to incorporate forms of mitigation or emphasis, accounts, and other specifications that, in other formats, are conveyed by separate turn components (Heritage & Raymond, in press).

BACKGROUND
Because linguists and philosophers of language demonstrated that "there are strong variations in the degree of dependence of the reference of the linguistic expressions on the pragmatic context of their production" (Bar-Hillel, 1959, p. 359), indexicality has been defined as the broader domain of contextual dependency of language (Levinson, 2004, p. 97), and indexicals as the means whereby aspects of context are grammatically encoded in language (Levinson, 1983). Through these expressions, speakers establish a relationship between the object that is being referred to (referent) and the means used to do the reference (indexical). With regards to this, linguists further distinguish between two dimensions. The first is a "spatialist" dimension of indexicality (also called "deictic" or "exhophoric"), where indexical expressions point to the situation physically or perceptually external to the utterance (Hanks, 2005, pp. 195 -196). The second is an "intratextual" dimension (also called "anaphoric" or "endophoric"), establishing a reference to an object not physically accessible to speakers but only mentioned in prior talk (Cornish, 2008;Green, 2009;Hanks 2005). Whereas in the first case the indexical is used more as a pointing device (i.e., "That is my car," where "that" refers to an object that is physically present and proximal), in the second the indexical establishes a coreference with the fully named antecedent (i.e., "He lost his wallet. It was new," where "it" refers to the wallet). Our study focuses on this second coreferential dimension of indexicality, as embodied in interaction.
According to the theories illustrated above, indexicals draw their meaning from the antecedent full-form reference to which they refer. In one of his lectures on indicator terms, drawing from interaction in a group therapy session, Sacks (1992) observed that, in conversation, indexical terms are used massively, often in chains, and in a stable way across turns, without people having to assign a formulation to what they are referring (p. 519). He proposed that these terms constitute a specific referential methodology ("machinery"), alternative to formulations, and that people use it for particular interactional reasons, besides mere coreference. He argued that one of these reasons is that indexicals enable people to do the work of "invoking the sheer fact of the setting without the specification of the setting" (Sacks, 1992, p. 520). More recent studies of reference to persons, both in linguistics (for a review of these studies, see Kitzinger, Shaw, & Toerien 2012) and in conversation analysis (Schegloff, 1996), have confirmed that indexical expressions, such as pronouns, are indeed frequently used in first position, in the absence of a full-form reference as its antecedent, thus supporting Sacks' original observation.
Our analysis offers an account for this practice in apologies, both when indexicals straightforwardly point back to an antecedent full form and when it is missing. Although in the small set of apologies we analyzed indexicals include pronouns and time references, we focus here only on pronouns, for reasons of space limitations.

DATA
From a subset of the total corpus of apologies (as described in the Introduction to this issue), we identified eight cases of IAs distributed among six calls. Two factors seemed most salient in accounting for the choice of referring to an offense deictically rather than naming it. As we suggested earlier, the first concerns whether the apologizer is the one who initiated the course of action leading up to the apology. The second, closely related to the first, bears on how much the would-be apologizer displays having access to the virtual offense and its circumstances. These features, associated with the way in which the IA establishes the reference within prior talk, provide the ground for differentiating two groups. To simplify, and drawing on categories used to explain the mechanism of repair (Schegloff, Jefferson, & Sacks, 1977), we call self-initiated IAs those in which the apologizer also initiates the line of action that leads to the apology and other-initiated IAs those in which the apology is produced in response to a course of action initiated by the potential offended person.
As mentioned before, the examples are not evenly distributed in the two groups; there are only two cases of the first group (in two different calls). In both groups the apology is produced in relation to circumstances that the apologizer has mentioned or that actually emerged in prior interaction in proximity to the apology and to which the indexical reference points back. The second group contains six instances of IAs occurring in four calls, where, in many cases, the apologies are repeated and/or reformulated several times, also using other formats. Here, by contrast, the apology is produced within a course of action initiated by the potential offended person (usually the caller) to which the apologizer responds with an IA. In these examples, the indexical is not associated with a precise full formulation of the offense, but it reaches back to the whole prior course of action, establishing a reference with what said in prior talk.

SELF-INITIATED IAS: REFERRING TO A PRIOR NAMED OR ENACTED OFFENSE
In this section we show that the simple reference to a prior explicitly identified offense is associated with specific interactional circumstances. These are, first, that the person who offers the apology is also the initiator of the project that will lead up to the IA. In the first case of this set, the apology is built as (one of) the reasons for calling and refers to a state of affairs that is not accessible to the interlocutor but which the apologizer makes available to him or her by naming it right before the apology is produced. In the second case, the apology deals with circumstances that are accessible to both parties and disrupt the interaction. In both cases the apologizer is the caller. Second, by so doing, the apologizer displays having access to the offense and its circumstances.

Reference to a Full-Named Offense
We are already familiar with example (4), which is the only case in which the indexical establishes a straightforward coreference to a previously full formulation of the offense (anaphora). In what follows, we foreground the issue of the apologizer's initiative and ownership of the course of action that leads to the IA.
Emma produces an IA in line 13. Emma herself has laid the groundwork for this apology, in her formulation of the offense in line 10, to which the indexical "that" refers back. Thus, the apologizer has first raised the relevance and the reason for the apology and, subsequently, produced it. Furthermore, we observe that the whole sequence leading to the apology occurs soon after the speaker's identification in the opening of the call. This strengthens Emma's status as the initiator of the action by calling in order to apologize. In this context, the indexical is used to establish a direct coreference to a prior formulation of a state of affairs that is thus considered a potential offense.

Reference to Physically Present Circumstances (Offense) Directly Available to the Apologizer
The practice of referring back to a prior referent is not restricted to cases in which the offense is named, as in (4). In the example below, the offense is not invoked by one of the interlocutors but takes place during the conversation itself.
In the course of the call, at lines 1-3, William asks Gordon to "hold on" while he deals with children disturbing the conversation. When he gets back to Gordon (line 17), he produces an IA. This reference is treated as unproblematic by both speakers, who promptly resume their conversation. What is particularly relevant here is the fact that, unlike the previous example, both participants have full access to the circumstances that constitute a possible offense. Thus, the apologizer does not need here to name the state of affairs for which to apologize. As in the previous example, the apologizer initiates the course of action whose upshot is the IA.

OTHER-INITIATED IAS: REFERRING TO A PRIOR UNNAMED OFFENSE
In four calls of our collection the apologizer expresses regret without having initiated the sequence that has generated the apology. This circumstance is associated with specific contextual features. First, apologizers are the called parties; they are the recipients of the telephone calls. 3 Second, and related to this, apologizers might not be (fully) aware of their being blamable and, also, not prepared to have to deal with this circumstance. Third, the apology is designed as responsive to the caller's prior action. In these cases the offense is neither expressly formulated nor does it arise from some contingent event; it is only invoked by the caller, thereby making relevant the production of an apology. In this way the offense has been conjointly constructed through the progression of talk, so that the indexical reference is endogenous to the interaction. In example (3), shown again here as (6), something similar happens.
Joyce's negative answer treats Leslie's question as a simple query, not worth an apology (on which see Drew & Hepburn, in press). Leslie's question is treated as a request for information; the minimal delay before Joyce responds (line 3 and the inbreath in line 4) are congruent indications of dispreference (Pomerantz, 1984; but see Kendrick & Torreira, 2015). The first signs of misalignment start to appear in Leslie's receipt (line 5). By displaying acceptance ("O " kay then:,"), 4 rather than a mere acknowledgment, Leslie conveys that her question had other ends. In addition, Leslie's turn presents contrasting features: It is designed to initiate a "possible pre-closing" (Schegloff & Sacks, 1973;Wright, 2011), but the final slightly raising intonation conveys a sense of noncompletion, as though she were expecting from Joyce some further elaboration. 5 Although facing Joyce's "no," Leslie, nevertheless, withholds the closing of the call or any move to a next topic (see the gap in line 7). It is only at this point that Joyce produces her apology. The delay with which the apology is produced indexes that only now has Joyce become aware of the consequences of her negative response and produces the remedial action. The apology is produced as the result of a pursuit, and it responds to the suspension of the progressivity of the interaction.
The next example illustrates another case in which the indexical does not refer to a prior full formulation but reaches back to prior interaction, from which it draws the sense of the offense as its meaning. In the conversation between Elaine and Miss Davids, Elaine's son, Elliot, wants to buy Miss Davids' house. Miss Davids has called back to report whether Elliot can deal privately with her or through the estate agents ("they" in line 1). The extract below occurs at the end of the call. In the preceding talk, Miss Davids has informed Elaine that, according to the estate agents, private negotiations must stop. In line 1 Miss Davids reiterates her position for the last time before the closure of the call.
In contrast to the prior example, here the apologizer is also the caller, bearing bad news. In this circumstance Miss Davids might be regarded as having initiated the course of action leading to an IA, in which case this example might seem to be a deviant case. However, to better understand the range of Miss Davids' initiative in relation to the virtual offense, two observations are relevant. First, from the opening of the call (not reported here for space limitations) we know that Miss Davids is returning the call after several previous attempts by Elaine to get in touch with her. Second, on several occasions during the course of the conversation (unseen transcript), Miss Davids denies responsibility for the problem ("I'll haf to ecce:pt Mann'n Co's a:rgument," "I'm really lef' between th'devil'n deep blue sea: I have no ohption," "I'm jus' lef' with no alternative."), and on a couple of occasions, she also apologizes ("yohr son will haftuh negotiate vIa Craft and Co 'm sorry t'sa:y,", "But that will haf to be involved with yo' son I'm afraid). In these circumstances Miss Davids is representing herself in the position where an apology is due, but for something for which she is not responsible. This intricacy, in which the sense of regret blends into condolence, blurring the nature of the apology, is resolved with the use of the indexical reference without having formulated the offense precisely (i.e, without naming it). In (7) the IA (line 10) is produced when the call is about to close, as indicated by the "possible pre-closings" (Schegloff & Sacks, 1973) in lines 3-9; it is deployed as the result of a pursuit, similar to (6). In response to line 1, Elaine produces minimal acknowledgement tokens (lines 2 and 4), withholding acceptance of the informing, and suspending the closure of the sequence until line 7, when she produces " " Okay." Thus, as in the previous example, the IA is deployed after the sequence has been closed with some hesitancy and after a pause (line 8). Miss Davids further delays the IA by means of the prefacing 'well.' Thus, similarly to Joyce in (6), Miss Davids seems to apologize because of the prior suspension of the talk's progression. In (6), the delay with which Leslie moves to the next topic seems to make Joyce realize that an apology is due; similarly, here, the apology wraps up the whole conversation, whose closure is delayed by Elaine's reaction. In both cases, in this terminal phase of the calls, the indexical is understood to reach back to all that has been dealt with in prior talk, 6 in (7) bearing bad news, having to stop private negotiation, additional fees for the estate agents, and so on rather than to a single formulation of the offense. Miss Davids designs an apology that, together with her regret, also conveys nuances of complaint, self-regret, and condolence.
Thus, by means of the indexical that refers to an unnamed offense, speakers manage to fine-tune their expressions of regret with the sense of the offense conveyed by the recipient in the prior interaction and by managing to preserve their own stance about it as it gets shaped in the ongoing interaction. Example (7) shows also that the use of indexicals, without a prior named offense, enables speakers to offer apologies in situations where they seem to consider themselves as not responsible for the offense.
In what follows, we see another complex situation in which the apologizer has not initiated the course of action leading up to the apology. In this call the speaker manages to mitigate and even elide his responsibility for the offense, even while apologizing (see also Drew & Hepburn, in press; Heritage & Raymond, in press). The night before this call, Gordon phoned his girlfriend Dana very late, thereby fueling Dana's mother's anxiety about her daughter who was not yet home. Dana is the caller, and our target turn is line 36.
The apology begins with the time-reference apology offered by Gordon at line 11. But, although he does initiate the sequence that leads up to the IA in 36, it can be observed that this is triggered by Dana's response in line 8, which Gordon repeats before apologizing. By saying that she is not okay, Dana seems to allude to what occurred the night before and to Gordon's failure to call earlier that morning (to apologize?). Thus, Gordon's apology in line 11 occurs after Dana has alluded to the trouble he has caused and thereby made relevant an apology; in this respect she, Dana, and not Gordon has initiated the sequence leading to the apology. In addition, Dana informs Gordon that she lied to her mother to explain his having called, thus aggravating Gordon's wrongdoing (Heritage & Raymond, in press).
However, in line 36, Gordon does not endorse this position. After agreeing ("No") with Dana's hypothetical scenario ("if I'd said (·) Gordy rang up to see if ¼ I'd come over tomorrow it wouldn't'v: gone down too well."), he prefaces his IA with a description of Dana's mother's reaction ("I didn't think she 'd uh (·) she'd panic"). In this way, he manages to foreground Dana's mother reaction as excessive, overshadowing his responsibility. The example shows that indexicals can be used to accommodate the speaker's goals to the interactional constraints, enabling them to apologize and, simultaneously, deny being culpable.
However, IAs are not associated only with a delayed realization of the offense or with reluctant apologizers. In the last example of this set, a series of apologies are offered for an unnamed offense. Leslie is calling to check whether Myrna will attend the meeting (see (6) above for Leslie's call to someone else about the same meeting). Having access to two calls by the same caller, made for the same reason but to different recipients, who both decline using an IA, is a particular fortunate coincidence, providing interesting insights into the IAs' formation and functions. The two different outcomes of the same request show that IAs are flexible resources, serving speakers' interactional needs when apologizing.
Example (9) starts with Leslie's request (the opening of the call is not shown). Myrna immediately apologizes in lines 2 and 11. In common with the previous examples, the apologizer is not the initiator of the course of action leading to the apology, but she finds herself "at fault" by virtue of Leslie's enquiry. The IA we are focusing on here is in line 22.
In the talk preceding our target line, Myrna frames a detailed description of the unfortunate circumstances that prevent her attendance to the meeting with the two apologies (lines 2 and 11), in which she provides accounts for the virtual offense. Myrna's behavior contrasts dramatically with Joyce's negative answer and subsequent apology in (4), which are brief, straightforward, and produced in a descriptive void of the circumstances of the declination. As Joyce's talk is laconic and brief, Myrna's is elaborated and informatively rich. Whereas Joyce first treated Leslie's question as simply an enquiry, Myrna right away orients to the implicit offense, either by not going to the meeting and/or because she has not phoned earlier to let Leslie know. 7 Myrna doesn't explicitly answer Leslie's opening enquiry until line 27 ("But I don't think I can Leslie,"); Leslie and Myrna nevertheless clearly orient to the declination from the beginning of their conversation, 8 as indicated by Leslie's absolutions in lines 12 and 23 (Robinson, 2004). However, when Myrna finally produces the IA in line 22, "it" can refer not merely to the declination of the request (though not formulated yet) but to everything in her extensive description of the problematic circumstances related to having to decline the request. Myrna's apology, thus inflated with additional meanings (including accounts), blends in with complaint and self-regret, adding to the mitigation of her responsibility for the offense.

CONCLUSIONS
Based on an investigation of a collection of apologies using indexicals to refer to the offense, we demonstrated that the function of indexicals in this context is not only that of establishing a simple coreference with a prior formulation of the offense but also that of providing a mechanism that enables speakers to deal with apologies in delicate circumstances, as when their responsibility for the offense is uncertain. As we all have experienced, people might find themselves at fault unexpectedly, having to apologize without accepting fault and responsibility for the offense or even without having previously been aware of having committed any offense; they are responding to interactional contingencies as they arise in the talk. In these cases, the would-be apologizer ends up expressing regret, without having initiated the course of action that has generated the apology (otherinitiated apologies). These cases cover most of the examples of our collection. Under these circumstances, the use of indexicals is associated with the absence of any prior explicit formulation of the offense. Because of their semantic emptiness, and despite the limited range of indexical expressions used, speakers exploit the position of the IA as a source of meaning. In this context, the indexical can be used to refer back to all that is being dealt with in prior talk, enabling the apologizer to tailor the apology to the sense of the offense that arises from the interaction. By contrast, when speakers initiate the apology (self-initiated apology), indexicals are used to establish straightforward anaphoric reference to a prior formulation of the offense. The analysis provides insights into the use of indexicals as an alternative "machinery" to formulations (Sacks, 1992, p. 520).