Over-Exposed Self-Correction: Practices for Managing Competence and Morality

ABSTRACT

When repairing a problem in their talk, speakers sometimes do more than simply correct an error, extending the self-correction segment to comment on, repeat, apologize, and/or reject the error. We call this “over-exposed self-correction.” In over-exposing the error, speakers may manage (and reflexively construct) a range of attributional troubles that it has raised. We discuss how over-exposed self-correction can be used to: (a) remediate errors that might suggest the speaker’s incompetence; and (b) redress errors that may be heard as revealing relational “evils” (implicating inadequate other-attentiveness) or societal “evils” (conveying problematic social attitudes and prejudices). The article thus shows how conversation analytic work on repair can provide a platform for studying the emergence and management of socially and relationally charged issues in interaction. The data come from a diverse corpus of talk-in-interaction in American, British, and Australian English.

Errors—speech errors, (Freudian) slips, slips of the tongue, gaffes, etc.—and mechanisms of their remediation have been the subject of intense cross-disciplinary interest due to their potential to throw apparent light on speakers’ cognitions, psychological processes, linguistic competencies, and sociological orders (e.g., Bolinger, 1953; Freud, 1904/1975; Fromkin, 1973; Goffman, 1981). In conversation analysis, errors and their correction are examined through the lens of repair, a set of practices—available to speakers (self) and recipients (others)—for dealing with problems of speaking, hearing, and understanding (Schegloff et al., 1977). Self-repair, whereby a speaker repairs his/her own talk, has been shown to provide evidence for the normative character of constructing social actions by bringing to the interactional surface the work it takes to design a turn (Drew et al., 2013; Jefferson, 1974, 1987; Schegloff et al., 1977). In doing self-repair, the speaker breaks the progressivity of the turn at talk, thereby exposing a bit of talk (the trouble source or repairable) as having been in some way inadequate or erroneous.¹ In this article, we examine a set of repair practices that we call over-exposed self-correction—repairs that do more than simply correct an error (i.e., repair simpliciter).

A rich tradition of conversation analytic research on self-correction and repair has documented a multitude of practices through which speakers fix problems in their talk (e.g., Bolden et al., 2012; Clayman & Raymond, 2021; Couper-Kuhlen & Thompson, 2005; Drew et al., 2013; Fox et al., 1996; Hayashi et al., 2013; Jackson & Jones, 2013; Jefferson, 1974, 2018; Lerner et al., 2012; Lerner & Kitzinger, 2015, 2019; Mandelbaum, 2016; Schegloff, 2013). As several of these studies observed (e.g., Jefferson, 1974, 2018; Mandelbaum, 2016), these practices differ in the degree to which an error—and its correction—is drawn attention to or exposed. On the one end of the spectrum is embedded self-correction, whereby an error is corrected without breaking the progressivity of the turn (Jefferson, 1987; Mandelbaum, 2016). For example, speakers may attempt to cover up an error by producing an and-prefaced repair solution, as in Extract 1, taken from a speech President Trump (mis)read from a teleprompter:

Extract 1: Furniture and future

Here the correction (from “furniture:” to “future”; line 6) is embedded into a turn-in-progress without a break in progressivity by incorporating it into a list (Jefferson, 1990, 2018). In this way, the speaker “recasts a possibly problematic utterance component as now having been merely the first item of a list” (Lerner, 1994, p. 25).

In contrast to embedded correction, exposed self-correction practices draw attention to the error in the process of correcting it. The degree of exposure depends on how much of the trouble source is articulated before the repair is initiated (what Jefferson, 1974 calls the “degree of error”) and on the particular repair practices deployed to fix it. For instance, in the following phone call (recorded in the 1960s), the speaker, in formulating a (racial) person reference, only articulates the first sound of what is hearable as the beginning of the word “colored” before cutting it off (“k-”) and replacing it with “nigro” (line 6; Jefferson, 1974, p. 193).

Extract 2: [Trio II] (adapted from Jefferson, 1974)

By partially verbalizing and then quickly replacing the problematic reference “colored” with, presumably, the less problematic (for that historic period) “nigro,” the speaker “achiev[es] a ‘just in time’ correction so that no supportable complaint can be made by its recipient” (Jefferson, 1974, p. 193).

Trouble sources are more exposed in the following cases, where the repair is initiated only after the trouble source is fully articulated. For example, in Extract 3, the cutoff after “people-” initiates a replacement repair (to “women” in line 2), and in Extract 4, an in-breath (.hh) after the person reference in the form of last name (“Durbin” in line 2) initiates an insertion repair to add an honorific (“>Senator< Durbin”).

Extract 3: [BCC] (Lerner et al., 2012)

Extract 4: Ford-Kavanugh Senate hearing [OC #51]

Further exposure may be done when the repair segment is extended beyond its minimal form, the repair simpliciter, by, for instance, rejecting the trouble source (Extract 5) and commenting on, repeating, and laughing about the error (Extract 6):

Extract 5: Old people [OC#47]

Extract 6: [GTS V:29] (Jefferson, 1987)

Practices like these over-expose the trouble source: In correcting the error, speakers halt the progressivity of the unfolding turn, beyond what would have been strictly necessary to fix the error, to do additional interactional work, such as repeat, reject, comment on, or apologize for the error. It is instances like these (further detailed in the “Practices” section) that we call over-exposed self-correction.

The continuum of error correction practices, with their differential potential to draw attention to or conceal the error, suggests that they may be marshaled to achieve particular interactional ends. Prior research has shown, for example, that embedded self-correction is particularly useful for removing inapposite hearings delicately (Mandelbaum, 2016), and the correction of minimally exposed errors (as in Extract 2 above) may subtly acknowledge the speaker’s and recipient’s identities (Jefferson, 1974). Furthermore, Jefferson (1996) shows how some “sound formed errors” (p. 9) can be treated as accountable when they are related to topics such as sex, race, fear, and hostility. Relatedly, focusing on slips associated solely with race in a small collection of public/media talk, Burford-Rice and Augoustinos (2018) noted that they are often accompanied by elaborate apologies, at the moment of production and/or at a later date; they observed that accounts within such apologies tend to deny that the slips revealed speakers’ racially prejudiced thoughts or intentions.

This article contributes to this body of research by exploring over-exposed correction practices. We argue that in over-exposing the trouble source, speakers do more than simply correct it. Rather, they manage (or reflexively reconstruct) a range of troubles that the error has raised. We have identified two distinct—and sometimes overlapping—issues that are managed through over-exposure: (in)competence and the potential for the error to be treated as revealing the speaker’s poor character, prejudices, etc. (what we have loosely called ascriptions of immorality or “social evil”).² This analysis thus uses the study of self-correction as a platform for understanding the way “isms” and similar problem actions and identities are managed, contributing to the growing interactional literature on these topics and, more broadly, on morality and accountability in interaction (e.g., Augoustinos & Every, 2010; Barnes et al., 2001; Edwards, 2004; Glenn, 2003; Rafaely, 2021; Robinson, 2016; Romaniuk, 2016; Speer, 2015; Speer & Potter, 2000; Stokoe & Edwards, 2007; Whitehead, 2017).

The article is organized as follows. After an introduction to the data and methods used in the study, we briefly overview the practices of over-exposure. In the remainder of the article, we discuss the interactional accomplishments of over-exposed correction practices: first, in managing errors that might suggest incompetence and second, in managing relationally or socially “evil” errors, i.e., errors that might threaten relationships and convey prejudices.

Data and methods

Using conversation analytic methodology (Sidnell & Stivers, 2013), we examine a diverse corpus of talk-in-interaction in American/British/Australian English that draws both on conversational materials (over the telephone and face-to-face) and a range of different types of institutional interactions (such as news and entertainment media programs and political events). The data come from “classic” CA materials, recordings collected by the authors, and public sources. Except for the data from the public domain, participants’ informed consent to use and publish the data was obtained and all transcripts anonymized. Our collection consists of approximately 100 cases of over-exposed self-correction (as defined in the introduction and detailed in the “Practices” section). For each case, we analyzed the repair practices through which an error is (over-)exposed as well as the interactional work these repair practices are employed to do in their local interactional contexts.

Precisely what counts as “over-exposed” is something we address in our analysis and discussion, but in making decisions about what belongs to the collection, we found it helpful to explore a number of “boundary cases” (Schegloff, 1997). For example, a trouble source may be minimally exposed via regular repair practices (as illustrated in Extracts 2–4) but may be accompanied by a variety of visible embodied actions (e.g., eye rolls, headshakes, frowning, etc.) that draw additional attention to the error. We have not included such examples in our analysis, but they provide a topic for future research.

Practices for over-exposing the error

In this section, we outline practices for implementing over-exposed self-correction—that is, practices that extend the repair segment beyond its simple form. The deployment of one (or several) of these practices is criterial for identifying our target phenomenon, i.e., for distinguishing over-exposed self-correction from repair simpliciter. These practices may be deployed before the repair is resolved, that is, prior to the repair solution (as in Extract 7) and/or after the repair is resolved, typically within the same turn (as in Extract 9). These practices include repeating the trouble source, commenting on the error, apologizing for the error, and rejecting the trouble source. Additionally, most cases have embodied components, such as laugh tokens, prosodically marked delivery, and for face-to-face interaction, distinct facial expressions, body movements, etc. We illustrate these practices with short fragments, before explicating the interactional work accomplished via over-exposed self-correction in subsequent sections.

Repeating the trouble source

Over-exposed self-correction may involve repetition of the trouble source (i.e., the problematic formulation), which highlights the error before providing a repair solution (cf. Jefferson, 2007). For example, in Extract 7, the speaker repeats the trouble source “fun” before replacing it with the repair solution “interesting.” The repetition is done with characteristic features of prosodic delivery (lowered pitch, “↓fun”), and the repair solution contains elements of laughter: “smile voice” and additional aspiration.

Extract 7: Fun interesting [OC #37]

For other illustrations of this over-exposure practice, see Extracts 11 and 18.

Rejecting the trouble source (“Not X, Y”)

Another common practice for exposure involves rejecting the error before correcting it using the “not X, Y” format. By explicitly rejecting the trouble source (“not X”), the speaker disallows or retracts it as a mistake, and the subsequent repair solution (“Y”) is offered as its correction (Lerner & Kitzinger, 2019, p. 8). For instance, in Extract 8, when discussing her friend James, Tammy refers to him as her roommate and then immediately rejects this trouble source by using the “not X, Y” format (line 1–2). In addition to the verbal exposure, the speaker also deploys lowered pitch and speeded up prosodic delivery.

Extract 8: Roommate [OC #13]

For an other example of rejecting the trouble source, see Extract 15.

Apologizing for the error

Schegloff (2013) noted that “sorry” can open the repair space, where it indexes a break in the progressivity of the turn and/or sequence. Exposure of the error may be done with an apology (using a variety of explicit apology formats), reflexively constructing the error as a transgression in need of remediation (Heritage & Raymond, 2016; Robinson, 2006). In Extract 9, the speaker, a news anchor, mispronounces the word “labor” as “labia” (line 4), producing what Jefferson (1996) called a “sound-formed” error (likely anticipating ia of “Kezia,” line 4), and following the replacement with a repair solution (“Labour”), apologizes for the mistake (line 2) while embodying something like confusion:

Extract 9: Scottish Labor [OC #56]

Fatigante et al. (2016) note that more expanded apology formats (such as “I’m sorry” vs. “sorry”) are used in environments dedicated to apologizing to launch expanded and elaborate apology sequences that may include explanations, accounts, recycling, and modifiers that intensify the speaker’s remorse (see also, Burford-Rice & Augoustinos, 2018). For an additional example of unexpanded (but repeated) apology, see Extract 16; for an instance of a more elaborate apology, see Extract 17.

Commenting on the error

Commentary on the trouble source encompasses various activities such as accounts, disclaimers, and self-reproach. The producer of the trouble source may comment on their error either before or after they provide a repair solution. In Extract 10, the trouble source is the speaker’s gesture of “air quotes” (during line 4). The commentary includes self-reproach (lines 5–8) deployed to manage the gestural error.

Extract 10: Joycons [OC #99]

For another example of repair commentary, see Extract 14. Sometimes commenting and apologizing are combined, as in Extract 19.

In this section we have outlined four practices through which speakers extend repair segments and over-expose trouble sources in self-initiated self-repairs. In what follows, we explore in more detail the variety of activities that these repairs are involved in.

What gets accomplished through over-exposed correction

When self-correction is carried out via repair practices, the repair practices necessarily expose a trouble source (as outlined in the introduction). In over-exposing a trouble source, speakers reflexively construct the existence of a trouble that can’t be fixed by repair simpliciter. In this section, we explore different kinds of anticipated, actual, or incipient troubles speakers are managing through the practices of overexposure outlined previously. We have divided our cases into those that involve managing potential ascriptions of incompetence and those concerned with immorality. Although useful, the distinction might not always be fully sustainable as participants may—and sometimes do (e.g., Extract 12)—treat incompetence as a kind of moral failing (cf. Bolden, 2018; Garfinkel, 1967).

Managing potential ascriptions of incompetence

In this section, we explore instances where the error may be taken to reveal the speaker’s lapse in competence (in a particular domain), which the over-exposure is then occupied with fixing. These kinds of errors involve (potentially) silly or funny mispronunciations or inapposite lexical substitutions. Our analysis shows that a key task for speakers is to manage their error in a way that treats it as an unintentional slip (Jefferson, 2007, pp. 451–452).

Extract 11 shows how a speaker manages being heard as incompetent by over-exposing their speech error for comedic effect. The segment is from a televised talk show; here American pop singer Ariana Grande (AG) tells a “funny” story (lines 1 and 7) about making a music video with another artist.

Extract 11: This take [OC#38]

In line 9, AG mispronounces the word “take” as “twake,” cutting it off to initiate repair. (This sound-formed error is likely due to anticipating the sound from “twerking” (Fromkin, 1973; Jefferson, 1996).) Before providing a repair solution (“take” in line 12), AG repeats the mispronounced word, preframed with “this,” fully articulating it with an elongated vowel sound and questioning intonation (“twa:ke?” in line 10), as she leans toward the host and points toward her mouth. All of this over-exposes the error. AG’s use of questioning intonation, along with the pointing gesture, convey that she is interrogating herself, displaying that what she just said was not intentionally produced. AG thereby highlights the error for self-teasing and playful effect, which her addressee treats as laughable (line 11). One important institutional task for both participants is to entertain the audience. Treating her misspeaking as a laughable via the over-exposure (rather than, say apologizing for it or just fixing the error via repair simpliciter) fulfills that goal.

Extract 12 illustrates the use of different—and more expansive—over-exposure practices: a token of apology (“sorry”) followed by a self-interrogating commentary. The extract is taken from a video-chat conversation between two sisters. In line 3, Jean launches a new sequence—a request for time—possibly to find out whether the call should be brought to a close.

Extract 12: Eleven [OC#60]

In line 5 Lynn cuts off her initial time formulation (“ele:v-”) while directing her gaze presumably to a time source on her desk. Lyn registers her error with “Oh,” displaying “news to consciousness” (Heritage, 1984), thereby casting her mistake as not consciously produced. Her apology token and self-interrogation (“>Why’d I say eleven.<” in line 6) extend the repair space further. The latter treats what she has just said as puzzling, reflexively recasting the error as a slipup (and herself as a competent teller of the time). Hence in over-exposing her error, Lynn treats the aborted production of “eleven” as opening herself up to being seen as incompetent—as incapable of getting simple, obvious information right when requested to provide it.

The two segments discussed in this section illustrate several practices speakers can use to manage errors that open them to attributions of incompetence or stupidity. By repeating the error before correcting it (as in Extract 11)—a common over-exposure practice for this sort of error—the speaker recasts the error as a silly (and possibly laughable) slip of the tongue (Jefferson, 2007). On the other hand, by apologizing and interrogating their production of the error (as in Extract 12), the speaker treats herself as accountable for having made it.

Managing potential ascriptions of immorality

A major environment where self-correction is over-exposed is where the speaker treats his or her error as potentially displaying some kind of “evil” or a moral transgression. Such evils can include relational evils, where the error could show lack of care and attention to others, or societal evils, involving ageism, racism, sexism, political self-interest, or the use of taboo language, among others. Our analysis shows that practices of over-exposure that manage such errors include (more or less expanded) apologies, which treat such errors as transgressions, rejections of the trouble source (in the “Not X, Y” format), accountings, and self-admonishments, often in combination with embodiments of embarrassment, laughter, and the like.

Relational evils: Managing errors that threaten personal relationships

Some errors may have problematic interpersonal implications, such as a failure to keep up with one’s relational or epistemic obligations. We find that in these cases, over-exposed correction is deployed to remediate or counteract these implications, suggesting that repair simpliciter, merely fixing the error, is not sufficient to right some relational wrongs. Practices for dealing with such interpersonally charged errors include apologies and enactments of shock or remorse, among others.

In Extract 13, the speaker over-exposes her error when she mispronounces the name of her coworker in the course of thanking her publicly. Here, Natalie Portman (NP) is giving an Academy Awards acceptance speech. While listing the people who helped her prepare for the role to thank them, she displays difficulty in articulating the last name of one of the people she is thanking and ends up mispronouncing it (lines 4–5).

Extract 13: Kostritzy [#OC65]

After articulating the first name “Olga” (line 4), NP launches into a word search for Olga’s last name (line 4–5). In line 5, NP resolves the search with a version of the last name (“Krukritzky-” in line 5), which she cuts off and immediately replaces with “Kostritzky” (line 5). Although this replacement repair is sufficient to fix the apparent mispronunciation, the repair segment is extended with an apology (“sorry,” line 5). Simultaneously with the unfolding word search and self-correction, NP enacts “being flustered” (smiling, gesturing, closing her eyes, etc. during line 5), further amplifying the error and its remediation. Through these over-exposure practices, and particularly via the addition of “sorry” following the repair solution, she treats the error not as a simple mistake but as a transgression in need of a remedy (Fatigante et al., 2016; Heritage & Raymond, 2016; Heritage et al., 2019; Robinson, 2004).

How this error is managed is shaped by aspects of this institutional context (a large live televised audience; the speaker is a movie star, and the referenced person is a behind-the-scene colleague) and by the course of action underway (thanking others during an acceptance speech)—all of these having a potential to aggravate the “evil” implications of the mispronunciation. By doing self-correction in this over-exposed way, the speaker conveys her accountability for knowing and being able to accurately pronounce the name of the person she is thanking, treating the mispronunciation as potentially conveying her improper or insufficient other-attentiveness.

Extract 14 illustrates how over-exposed correction may be deployed in the context of personal relationships. In this segment, taken from a mealtime interaction between three friends, in the course of explaining how they access cable in their apartment (data not shown), Lola refers to somebody with the nonrecognitional reference “Sel’s ex-boyfriend” (line 3) (Sacks & Schegloff, 1979). As it turns out, this reference is inaccurate as the relationship (between Sel and the boyfriend) is apparently still ongoing.

Extract 14: Ex boyfriend [OC12]

In line 3, Lola corrects the problematic person reference (“ = I mean not ex-boyfriend.”) by negating the error with “not” while shaking her head (Lerner & Kitzinger, 2019). She then goes on to further expose the error by treating it as particularly egregious with an expletive “Dhammit” (line 4) and a self-interrogation via a why-interrogative that takes a critical stance toward it: “Why’d I say that.” (line 4; Bolden & Robinson, 2011). In the extended repair segment that follows (lines 5–16), the speaker and the other participants deal with the error in various ways: for example, by attempting to minimize the damage it might cause the couple (by knocking on wood; lines 7–8), by presenting the error as shocking (“Oh my god” in line 11 and again in line 16), and by expressing remorse for making it (“<I feel so bad”; line 16). This extended repair segment is brought to closure when Lola produces a recognitional person reference “James” (line 16) and returns to her telling (line 18).

This extract illustrates how participants deploy practices of over-exposed self-correction to manage negative attributions about them as, for example, “poor friends.” Referring to a friends’ relational status in an egregiously inaccurate way may convey that the speaker is not appropriately other-attentive, having failed to properly track her friends’ relationship (Morrison, 1997). Further, by inaccurately treating a romantic relationship as having ended, the speaker might be seen as anticipating its termination—something one should not wish on others. Lola’s “Let’s all knock on wood” (line 8) may also indicate a concern that characterizing the boyfriend as an “ex” may bring bad luck by jinxing the relationship. The extensive over-exposure this error receives also suggests that the degree of exposure (i.e., from minimally to extensively exposed) may be proportional to the egregiousness of the error: When the interactional stakes are high, the speaker may do further work to repair the damage, thereby reflexively constructing the error as particularly grave.

Societal evils: Managing errors that may reveal speakers’ category prejudices

In the previous section we documented how getting a name wrong or a relationship status wrong could occasion elaboration over and above repair simpliciter to manage negative interpersonal attributions such errors may bring about. We will now consider repair that may be occasioned by the exposure of possible “societal evils” such as ageism, racism or sexism; by the production of transgressive or taboo sexual language; and by displays of inappropriate political self-interest. Again, the interest is in the “extra” interactional work that such errors occasion to manage moral implications of such transgressions.

In Extract 15, a CNN analyst Harry Enten (ENT), in the course of presenting recent polling numbers in the Democratic primary elections, refers to a survey sample of voters of “over fifty” as “old people” (line 6). Alisyn Camerota (CAM) and John Berman (BER) are cohosts.

Extract 15: Old people [OC#47]

In his excited exposition of data showing Biden’s increased lead in polling, Enten starts to underline the significance of the data, characterizing the relevant sample as “ol:d people” (line 5). He first retracts the reference “°↓not old people↓°” (lines 5–6) and then replaces it with “ol:der people”—the contrastive stress on the second syllable of “ol:der” locates the absence of the comparative suffix “er” as the problem being fixed. The trouble source is over-exposed in two ways. First, the negation of the trouble source (“not old people”) extends the repair segment beyond the repair simpliciter, projecting the repair solution as a correction (Lerner & Kitzinger, 2019). Second, in the 0.6-s pause after the repair solution, Enten produces an embodied stance combining an elaborate eye-roll (cf. Clift, 2021), facial turn, and gesture conveying something like exasperation or disbelief in what he has said. Only after this does Enten resume his suspended TCU with a further element (“c-come out …,” line 07).

The elements of over-exposure manage a specific kind of trouble. Referring to over-50s as “old people”—a stage-of-life category with which they may not self-identify—may be heard as potentially offensive or even ageist, which the comparative category older avoids. By going beyond mere replacement and inserting an explicit negation, Enten displays his attention to, and rejection of, this categorical usage, and in enacting exasperation with himself the speaker shows that he understands the egregious nature of his error.

This understanding of Enten’s words is developed in situ by the hosts Camerota and Berman, who collude in (jokingly) chastising Enten for his error after he has finished his report. In line 14, Camerota issues a “shaming interrogative” that draws attention to Enten’s prior words as the source of something problematic (Potter & Hepburn, 2020) and invites Berman to join in on admonishing Enten. Berman responds by proffering the term “old people” (line 13) as the transgression. Camerota and Berman both continue to chastise Enten for the error (lines 14–16) while laughing (lines 14 and 17–18) and smiling to modulate the admonishment, turning it into more of a tease. This post hoc treatment of the error by the coparticipants brings to light the exact attributions (of ageism) Enten’s self-correction attempted to head off and may be oriented to local institutional contingencies—the age demographics of the CNN’s target audience (leaning older) and of the program’s cohosts vis-à-vis Enten’s relative youth.

In Extract 16, presidential candidate Pete Buttigieg (PET) is talking about campaign finance reform with host Charlamagne tha God (CHA), in front of a predominantly African American audience. In line 2, Buttigieg is starting to list campaign finance problems:

Extract 16: Black money [OC#77]

Buttigieg names “black money” (line 2), which he subsequently repairs to “dark money” (line 3). This error is over-exposed in a number of ways. After a cutoff (on “black money-”) that initiates repair and a word search (line 3), Buttigieg prefaces the repair solution with an apology (“sorry” in line 3), constructing the error as a transgression. The error is further over-exposed when audience laughter (line 6) occasions further apology from Buttigieg (line 7). The feature of the error that requires over-exposure seems to be the association of “black” as in “black people” with “dark money.” Charlamagne highlights this in line 9, which playfully, yet pointedly, spells the mistake out as “African American money,” occasioning further laughter from his audience (line 13). Buttigieg further exposes his error by jokingly turning black money into a comment on donations from black voters (lines 12, 14).

The over-exposure of the correction (from “black money” to “dark money”) manages the potential for the error to suggest that Buttigieg has negative or even racist views of black people. It is notable that Buttigieg apologizes, twice. The proliferation of apologies may be occasioned by the fact that members of the defamed category are present both in the form of his interviewer and the audience.

The previous two segments revolve around what social scientists might gloss as issues of ageism or racism; the next involves what might be heard as an offensive and perhaps sexist slur. Extract 17 shows how over-exposed correction is used to manage the potential problematic inferences made available by a program host’s use of the term “bitch.” Here the host introduces a new feature of Fitbit Versa, a smart fitness tracking device, called “Fitbit Coach”—a name the host (H1) has trouble pronouncing correctly (from line 3).

Extract 17: Fitbit [OC#27]

In line 3, H1 initiates repair on the mispronunciation “fit bitch coa-” by cutting off “coa-.” (The mispronunciation is another instance of “sound-formed” error (Jefferson, 1996), anticipating “ch” in “coach.”) The speaker immediately over-exposes the error with “ohp- (.) oops-” and laughter (line 4), thereby producing it as an accidental misspeaking rather than an intentionally articulated expletive. However, in her subsequent attempt at a repair solution, H1 reproduces the same error “fit bitch” (line 4). After another attempt to advance a repair solution (with “<I mean-” in line 4), she halts the progressivity of her repair to comment on the error (lines 4–5). This over-exposure again casts the trouble source as unintentional, indicating surprise at her own conduct (“ohh m(h)y g(h)osh”). The lengthier over-exposure (compared to the more minimal “oops”) may be fitted to the egregiousness of repeating the same offensive word twice.

H1 finally produces the repair solution, again with interpolated laughter, in lines 5–6, stretching the production of “fitbit.” In slowing down the production of the repair solution, H1 demonstrates a deliberate attempt to avoid making the error for a third time. The extensive interpolated laughter throughout the over-exposure may manage the delicate task of producing an obscenity and show attentiveness to its problematic status (Jefferson, 1985). As Jefferson (1985) and Shaw et al. (2013) demonstrate, laughter can manage recipient responses by inviting recipient laughter or encouraging recipients—in this case H2 and the overhearing audience—to affiliate with and not take offense at H1’s trouble and her attempts to fix it. Indeed, following her third attempt to correct her error, H2 subtly congratulates H1’s more-or-less successful correction (“You got it,”; line 7) and then minimizes the impact of her error, absolving H1 of serious wrongdoing (“We ↑kne(h)w what you mea:nt,” line 9). H1 responds with a bare apology (“Sorry.” in line 10) and resumes her ongoing TCU.

In line 16, after completing the scripted monologue (interpolated with laughter at several points), H1 delivers another account for her error, pointing out the difficulty of saying “fitbit” and “coach” (“It’s those two words together.”). This account attributes the error to a pronunciation difficulty (suggesting a lapse in competence), possibly to head off more problematic ascriptions of immorality. The speaker follows up (on lines 21, 23, and 25) with a series of upgraded apologies. The expansion and recycling of the apology treats the error as particularly egregious and emphasizes the speaker’s responsibility and regret for having committed it (Fatigante et al., 2016; Heritage & Raymond, 2016).

If the previous three extracts display attention to “evils” that we can broadly gloss as ageism, racism, and sexism, in the next one the error points to political dissembling and inappropriate partisanship. In Extract 18, Mike Pence, then Governor of the state of Indiana, is being questioned about budget negotiations between the United States’ two leading political parties: the Republicans (Pence’s party) and the Democrats. The interviewer (INT) proposes a version of what the governor is doing as playing a winning move in a political game with the Democrats (line 1). In response, Pence rejects this characterization (line 3) and begins to formulate an alternative description of the strategy (lines 3–5):

Extract 18: American people [OC#50]

In his defense of the Republican’s negotiating strategy, Pence characterizes it as aiming to “score a victory: (0.2) for the Republican people.” (line 5), which can be seen to reinforce the partisan version of events offered by the interviewer (line 1). After a gap (line 6), in line 7, Pence initiates a replacement repair, starting to offer Americ(an people) but in the midst of producing this repair solution, he cuts himself off, laughs, and then repeats the trouble source “>the Republican peopl-<.” By interpolating the correction with laughter and then repeating the error, Pence over-exposes the inapposite initial construction, apparently in an attempt to present the error as accidental or “silly” (cf. Extract 11; see also Jefferson, 2007, pp. 451–452). In line 8, in another attempt to articulate the repair solution, he goes on to reproduce his entire turn constructional unit and then negates what his error revealed with an increment (Schegloff, 2016) “not for the Republican party” (line 9).

In this case, Pence’s problem is that in the context of emphatically countering a construction that he is playing partisan games, he appears to reveal that he is acting in an entirely partisan manner. The elaborate over-exposure—via repeating the trouble source with laughter and a full delivery of the correct form—highlights the significance Pence attaches to the attributional problem he must manage. This segment also suggests that there may be a preference for casting, whenever possible, an error as accidental (by repeating and laughing at it, for example) rather than something motivated by uncaring, immoral, or evil intent.

Errors may involve, and reveal, a multitude of sins. In the next segment we can see relational evils (getting someone’s name wrong), combined with inappropriate political partisanship (a BBC interviewer using an abusive slur for a government politician to another government politician), combined with potential sexism (using a common derogatory word for vagina), combined with the transgression of using a highly taboo word on public broadcasting. Extract 19 is from a televised political debate in which a BBC presenter, Victoria Derbyshire (VD), refers to the British Conservative politician Jeremy Hunt as “Jeremy Cunt” (line 2). Our focus is on how the presenter manages her error with over-exposed self-correction. In this case, the commentary accompanying the error is extensive, involving several rounds of apologies to manage the erroneous use of a vulgar and potentially offensive word.

Extract 19: Jeremy Hunt [OC#52]

Following the speaker’s production of the error (line 2), VD stops mid TCU to apologize for it (“I’m so sorry”) while putting her hand to her heart in a gesture of contrition, before producing the replacement repair solution “Jeremy Hunt” (line 3). The apology acknowledges her responsibility for her offensive behavior (Robinson, 2004) and, in doing so, over-exposes the error. The speaker treats the wrongdoing as worthy of a dedicated apology via a full TCU (“I’m so sorry”) rather than prioritizing progressivity with a simple “sorry” format (Fatigante et al., 2016). The more elaborate apology further exposes the error, drawing attention to its especially problematic nature as an unsayable.

Following the repair solution, the speaker could have continued with her turn; however, she instead further over-exposes the error by extending the repair segment to account for the error, addressing (via eye gaze) the other presenters in the newsroom (lines 4–5). She first presents herself as someone for whom this error is entirely uncharacteristic (“.hh I’ve never said that before in my li:fe,”; line 4), discounting the possible inference that she might habitually use such language. She then thematizes the gendered, possibly sexist, role of using this term (“it’s u:sually men who say that,” in line 5), perhaps alluding to others in broadcasting who have referred to Jeremy Hunt in this way and perhaps thereby suggesting it is an easy mistake to make. She then further over-exposes the error by producing two additional expanded apologies, the first upgraded with a recycled, repeated modifier “really” (“I really re:ally want to apolo#gi:ze.”; lines 5–6), followed again by the full “I’m sorry.” format (line 6)(see Fatigante et al., 2016). The repeated and expanded apologies seem to demonstrate the BBC presenter’s understanding of her error as particularly problematic (Heritage & Raymond, 2016), and it is only after these repeated apologies that the presenter returns to her abandoned TCU (line 7). By overexposing the correction in such an elaborated fashion and producing explicit, recycled, and modified verbal apologies, the speaker presents herself as cognizant of the inappropriate nature of the error in the context of a publicly broadcast BBC news program nationally broadcast and designed to provide unbiased and professional reporting of current affairs.

Overall, these segments point to the possibility that egregious errors require extraordinary measures to redress the problem—and, reflexively, practices for redressing the problem construct an error as more or less egregious. In each case, a simple replacement repair may have failed to manage the incipient problems. Additionally, speakers may strive, when possible, to (re)cast such errors as innocent misspeakings (as in Extracts 17 and 18), displaying a preference for coming across as incompetent rather than “evil.”

Discussion

This article has developed conversation analytic research on self-correction and repair, focusing on cases where self-correction is over-exposed, i.e., where speakers hold the progressivity of the turn beyond repair simpliciter, to apologize, comment, self-admonish, repeat, or reject the error. We have shown that ordinary repair operations (Schegloff, 2013), such as replacing one word for another, may not be enough to fix troubles in speaking. Instead, speakers becoming occupied with fixing the trouble that their trouble source has got them into, resulting in a repair segment that is expanded to include a range of verbal and embodied practices. We show that—and how—such practices manage errors that may be taken to suggest the speaker’s incompetence and/or bad character, including errors that may threaten the speaker’s personal relationships or are hearable as revealing the speaker’s categorical prejudices. Our findings thereby contribute to the growing research into how the machinery of repair can tackle issues beyond simply dealing with problems of speaking, hearing, and understanding.

Our study has interesting parallels with Goffman’s (1981) distinction between slips and gaffes. Goffman argues that slips—words that “have gotten mixed up, or mis-uttered”—should be “seen as a consequence of confused production, accident, carelessness, and one-time muffings—not as ignorance of official standards or underlying incompetence” (p. 209). In contrast, gaffes—“breaches in ‘manners’ or some norm of ‘good’ conduct”—indicate “ignorance of what one would have to know about the rights and biography of one’s coparticipants to conduct oneself with moral sensibility in regard to them” (pp. 210–211). Over-exposed correction practices documented here can be marshaled to address—and redress—errors of both incompetence and ignorance. Importantly, errors do not reveal ignorance or incompetence; rather we show that speakers orient to the errors’ potential for being hearable as revealing such underlying incompetence and moral insensibility—and these are precisely the incipient troubles being fixed through over-exposure practices.

Our analysis suggests that the more extreme the error, the more speakers engage in remediating the problem—and reflexively, by doing additional remediation, the speaker publicly calibrates its egregiousness. Interestingly, a large portion of our cases come from public settings, such as celebrity or news interviews. In these contexts, an overhearing audience may sensitize speakers to attributional matters, creating occasions with complex recipiency. For example, Enten’s use of the essentializing “old people” rather than the comparative “older people” (Extract 15) is likely to have an audience that includes many potential members of the category. Buttigieg’s use of “black money” instead of “dark money” (Extract 16), perhaps associating illegitimate financial practices with a social group, is delivered to an overwhelmingly black audience. The host’s use of “fit bitch” rather than “fitbit” (Extract 17) is streamed to an audience that is likely at least 50% female. In each of these cases, there are others directly present in host roles, adding complexity. And there may be specific institutional guidelines governing the reference to social categories, which may or may not be oriented to, adding yet more complexity.

In addition, we have shown the importance of recognizing the institutional task at hand and the course of action the speaker is engaged in. For example, if the goal is to entertain, treating the misspeaking as a laughable via the over-exposure fulfills that goal. Further, entertainment programs, such as a celebrity interviews, may have different progressivity constraints compared to a news program. In such contexts, elaborate error over-exposure may become an interactional resource for doing entertaining (as in Extract 11).

It is notable that self-corrections are regularly accompanied by nonvocal embodied conduct such as eyebrow raises (Extract 9), frowns (Extracts 9, 10, and 12), eye rolls (Extract 15), and headshakes (Extract 14). Semiotically these are conventional ways of marking attentiveness to having made an error that requires correcting; the headshakes, eye rolls, frowning, etc., display varying degrees of self-admonishment or exasperation with self. That is, they reflexively produce a speaker who is (now) not only catching and fixing an error but recognizes that this error is potentially egregious. Although this nonvocal conduct is both visible and understandable to recipients, it is not typically built for responding to. There is more systematic research to be done on the role of nonvocal conduct in relation to different forms of over-exposure, which might include a broader reconsideration of nonvocal elements accompanying self-repair that is not otherwise over-exposed.

By explicating over-exposed self-correction we throw light on trouble that goes beyond conversational errors to implicate possible relational problems (lack of other attentiveness, say) or incipient socially charged problems (racism, say). There has long been debate on if, and how, conversation analysis can contribute to the study of socially charged topics (e.g., Billig, 1999; Schegloff, 1999). This article demonstrates how a careful analysis of conversational repair can document participants’ orientation to—and treatment of—social evils.

The current study can be situated in the context of the existing interactional literature on the emergence and management of socially charged issues. Prior work has documented ways in which negative assessments of—and complaints about—individuals in contexts where their category membership is potentially salient may be constructed so as to avoid toxic implications for the speaker (Augoustinos & Every, 2010; Edwards, 2004; Potter & Wetherell, 1988; Whitehead, 2013) and how complaints can be built through reports of racial insults (Stokoe & Edwards, 2007). Another strand of work considers the way speakers respond to talk that may display racism, sexism, or other category-based prejudices (for reviews, see, e.g., Speer, 2015; Whitehead & Stokoe, 2015). This research has shown that stereotypic constructions may be challenged on the basis of their “inaccuracy” and/or moral shortcomings (Whitehead, 2018) and that responses to stereotypic claims may be designed to give speakers an opportunity to withdraw or back down (Whitehead, 2015). Furthermore, hearably racist generalizations can be undercut through repackaging them as extreme case formulations (Robles, 2015). Additionally, this work has identified repair and correction as important mechanisms for bringing to light and remediating “isms” in the talk of others (e.g., Land & Kitzinger, 2005 on heterosexism; Weatherall, 2015 on sexism).

The current study contributes to this literature by considering cases where self-repair is over-exposed when socially charged topics and categories, including issues of racism, sexism, and ageism, become potentially live in interactional settings. We show that such errors are managed using a mix of sometimes extended self-admonishments, accounts, and apologies, sometimes combined with variously positioned laughter and nonvocal conduct (cf. Burford-Rice & Augoustinos, 2018). One issue participants may manage is whether the error is revealing some underlying attitude or stereotype or whether it is an innocent slip. Participants hold up the progressivity to focus on this management, showing the importance of the task. They may or may not be successful in this management. Our analysis considers how it is done; success, or not, is a participants’ issue. And there is further complexity as the “isms” examples have multiple addressed recipients and overhearing audiences who may come to different judgments.

Our general point then is that repair can be a revealing site for considering the management of speakers’ orientations to their own in/competence and to charged societal issues. Clearly, further work needs to follow up the institutional specificity of these practices and to put energy into generating a broader range of mundane data from different settings and social groups. Following Stokoe and Edwards (2007), we can quote Zimmerman’s (2005) observation that the big issues of power, inequality, racism, and the like “have a humble home hidden in plain sight, in the ordinary workings of social life” (p. 445).

Acknowledgement

We would like to thank Gene Lerner, Anita Pomerantz, Jeffrey Robinson, audiences at conference at which earlier versions of the paper were presented, and the three anonymous referees for their insightful suggestions and feedback at various stages of this project.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ As Schegloff et al. (1977, p. 363) point out, what gets repaired is not necessarily an error, nor are all errors repaired. It is through the deployment of repair practices that the targeted talk may be constructed as erroneous in some specific way. The repair practices we discuss here (i.e., over-exposed self-correction) are designed to treat the trouble source as an error. For this reason, we refer to trouble sources as errors (when they are treated as such by the speaker) and to repair practices as correction.

² Some over-exposed correction is deployed to manage epistemic (un)certainty (see, e.g., Lerner et al., 2012); however, due to space limitations, we are not covering these usages here.