Social Interaction

1. Introduction

A rich tradition of conversation-analytic research on mealtime conversations shows the inextricable link between eating, drinking, and sociability. In fact, the multiactivity nature of conversing while eating and drinking renders it an ideal site for studying how participants coordinate a diverse range of multimodal resources, including talk, bodies, and objects to manage multiple, and sometimes competing, courses of actions. Naturally, eating and drinking inhibit speaking for physiological reasons. When liquid occupies the oral cavity, oral articulators are rendered inoperable (see Hoey, 2018). Chewing and swallowing block any airflow from lungs, which is essential for sound production (Ogden, 2023). Although participants can still "talk" with their mouth full, it is often seen as poor manners, and the range of sounds that can be produced is greatly limited (see Wiggins, 2010, for gustatory mmms). In this paper, we focus on moments when the trajectory of eating and drinking is brought to a halt to favor speaking precisely before liquid or food is passed into the mouth and show how such halts play a role in safeguarding contiguity in interaction.

Sacks (1987) describes the preference for contiguity between question and answer, e.g., one would typically answer the second question first to preserve that contiguity. We use the term "contiguity" to capture a related, but larger and more prevalent feature of interaction-the idea of being "on time," or placing the right item at the right time, which we observe at various levels of interactional organization. It is important, for example, to start on time at the possible completion point of a current TCU (Sacks, Schegloff & Jefferson, 1974), to initiate repair on time in the next turn vis a vis the trouble source (Schegloff, Sacks & Jefferson, 1977), and to place a topically relevant item on time in an unfolding conversation or else it might never get another entry point (Sacks, 1992). Such contiguity is also what Schegloff (2007) refers to as "adjacency or 'nextness'" that involves "moving from some element to a hearably-next-one with nothing intervening"-"the embodiment of, and the measure of, progressivity" (p. 14). Our interest is in demonstrating how the halt of eating and drinking is produced to ensure the timely production of what is due next so that its relevance and impact would not be delayed and weakened.

2. Background

2.1 Multiactivity and materiality

Multiactivity involves the organization of multiple activities "concurrently or serially" (Haddington et al, 2014, p. 5) in a setting characterized by the management of multiple temporalities (Mondada, 2014) that requires either intrapersonal or interpersonal coordination (Deppermann, 2014). Mondada (2014) distinguishes between three temporal orders of multiactivity: (1) "parallel order" where activities draw on different resources and thus can run in parallel, (2) "embedded orders" where activities are temporarily hierarchized as a result of constant mutual adjustments, and (3) "exclusive order" where one activity is put on hold in favor of the other (p. 35). Stefani and Horlacher (2018), for example, use hair salon interactions and driving lessons to show that professional work (i.e., cutting or styling hair) and chatting with a client can run in parallel smoothly when different multimodal resources are drawn upon (e.g., manual actions vs. talk). However, when talk is required for both professional work and chatting (i.e., giving driving instructions and chatting), participants may prioritize one over the other or come up with solutions to simultaneously juggle multiple actions. Similarly, Nishizaka and Sunaga's (2015) analysis of how volunteers simultaneously massage and converse with Japanese tsunami evacuees illustrates that talk can also be marshalled to facilitate massaging. Some have also described how participants manage trouble incurred by multiple involvements. Helisten (2019), for example, documents the use of disjunctively positioned problem-noticings in multiactivity settings. Despite their disjunctive nature, these noticings are nevertheless treated as legitimate actions by participants as they target interactional trouble that threatens the overall progressivity of concurrent activities. Vatanen & Haddington (2023) show that adults can suspend or deny a child's request with an account that draws upon the incompatibility between the ongoing activity and the requested activity.

In the context of managing eating, drinking, and speaking, as Erickson (1992) notes, talking and eating together is "a collective accomplishment across communication channels and modalities, and across the activity steams of various individuals in the scene" (p. 396). In his analysis of a family dinner, Erickson observes that one's reaching for food is synchronized with speech by other participants. A diner routinely puts food on the fork shortly before an utterance is completed and place the food-laden fork in their mouth right after its completion. Remarking on the synchrony of speech and dining activities, he notes, "[food to plate and food to mouth] sequences seemed to be coordinated not only with turn exchange points but with syntactic and prosodic junctures within turns, such as inter-clause boundaries, intra-clause boundaries, and boundaries between breath groups" (p. 394). Lastly, most relevant to our current focus, Hoey (2018) shows that despite the conflicting demands between drinking and speaking, a drinking participant selected to speak next can draw on vocalizations (e.g., "mm") or embodied resources (e.g., an eyebrow flash) to proceed with both activities in parallel. They may also adjust their drinking trajectory (e.g., acceleration or retraction) to accommodate speaking.

Integral to multiactivity is materiality such as "objects, artifacts, technologies, tools, documents, and shapes that participants can see, hear, touch and/or manipulate and use for accomplishing actions and activities in interaction" (Haddington, 2023, p. 1). In their landmark volume on interactions with objects, Nevile et al. (2014) demonstrate how objects can be used as a situated resource for managing "interactional demands and relevancies" and how they can be "formed, constituted and potentially altered through interaction" (p. 14). Richardon and Stokoe (2014), for example, show how different objects such as the cash till ("cash register" in American English) and tables in the bar become resources for accomplishing the service encounter in the bar. Weilenmann and Lymer (2014) describe how paper documents support journalistic work, where they make the distinction between "object-implicating interaction" (e.g., when gazing at a day-list contributes to a topic shift) and "object-focused interaction" (e.g., where a post-it note that gets passed on to the editor becomes the focus of interaction), entailing respectively "incidental" vs. "essential" involvements of objects (p. 333). As Mondada (2019) observes, "[o]bjects in interaction can be seen, sensed, and intractionally managed under many different aspects" (e.g., epistemic object, sensorial object, buyable object) and "actions and sequences of actions ... make relevant not only next possible actions" but also a host of other matters including "material objects" (p. 59).

In this paper, we continue this work on multiactivity and materiality by further exploring the orderly coordination of talk, food, and drinks -- what Raymond and Lerner (2014) describe as "the management of intersecting courses of action [that] poses inescapable in-the-moment options (and opportunities) for participants" (p. 229). Our emphasis is on one such option, namely, suspending the in-progress action of eating or drinking, much of which involves interrupting the movement of the cutlery and drinking vessel (as objects) towards the mouth.

2.2 Gesture phases

Insofar as this suspension of eating or drinking is a phase of a larger movement, our analytical focus may also be situated within the literature on gesture phases. Although our focal phenomenon resides in a largely non-gestural movement (e.g., plate-to-mouth), such a movement is similarly dissectible into several phases of manual movements (Kendon, 2004; McNeill, 2005): preparation, stroke, and retraction. Preparation refers to the phase of movement from home position (Sacks & Schegloff, 2002) leading up to the stroke; stroke, as Kendon describes, is "the phase of the excursion in which the movement dynamics of 'effort' and 'shape' are manifested with greatest clarity" (p. 112); retraction (or recovery) involves returning the gesticulating body part to a resting position (also see "home-away-home" in Sacks & Schegloff, 2002). Lerner & Raymond (2021) show how the preparation phase may be foreshortened and the manual action interrupted, causing body troubles that require various adjustments. In Mandalbaum and Lerner's (2023) study of "self-service Manual Action Pathway (MAP)," a diner can offer food to another at the preparation (e.g., reaching for food), focal action (e.g., transferring or a mealtime item), and return (e.g., putting down utensils and/or retracting arm) phase of serving oneself food. The temporal organization of eating and drinking is similarly phase-structured, as shown in Hoey's (2018) three-part structure that includes a preparation phrase (e.g., bringing the vessel to mouth or taking a small piece of food with utensils), a focal action phase (e.g., pouring liquid into mouth or placing food-laden utensils in mouth), and a return phase (i.e., reversing the vessel or removing utensil from mouth, hand/utensil back to home position). Our focal phenomenon, the halt of eating or drinking, occurs between the preparation and focal action phases in the cases of both eating and drinking.

2.3 Gesture holds

Also relevant to our interest in the suspension of movement is the scholarship on gesture holds. Gesture holds are often found in repair sequences. In Seo and Koshik's (2010) study on ESL conversational tutoring sessions, a particular type of gesture hold is used to initiate repair, where either a sharp head turn/head tilt to the side or a head poke forward with an upper-body movement forward is launched after a trouble source and held through the following turn(s) until the problem is resolved. In fact, based on data from Northern Italian, the Cha'palaam language of Ecuador, and Argentine Sign Language, Floyd et al. (2014) found that across the three unrelated languages and two modalities, when parties initiate repair with an accompanying gesture "hold," they would not disengage the position until the problem is resolved and sequence closed (also see Manrique and Enfield, 2015). Based on L2 French conversations, Skogmyr Marian and Pekarek Doehler (2022) describe gesture holds as part of a multimodal ensemble to both account for breaks in progressivity and signal a "solitary" search that preempts any recipient turn incursions. Likewise, freezing one's gesture in midair along with a thinking face and mid-distance look is part and parcel of accomplishing a solitary word search in Dressel's (2020) study based on French conversations.

Gesture holds also appear to address issues of intersubjectivity and recipient uptake. Sikveland and Ogden (2012) examine how participants in Norwegian conversations hold gestures beyond the end of a turn-at-talk as a way to handle issues of shared understanding to achieve and maintain intersubjectivity. Based on video-recordings of Swiss German Sign Language (DSGS) interaction within an institutional setting, Groeber and Pochon-Berger (2014) found that turn-final holds occur recurrently in turns that set a strong action projection (e.g., questions) and embody the current speaker's expectations regarding next actions. In Japanese conversations, as Cibulka (2015) shows, the prolonged hold is produced to pursuit recipient response. Investigating depictive gestures produced during the final components of the ongoing verbal TCU (turn-constructional unit in Sacks, Schegloff & Jefferson, 1974) and extended into the following turn transition space in Finnish conversation, Lilja and Piirainen-Marsha (2019) found that the depictive gestures elaborate the TCUs by providing additional information on the verbal content of the turn such as visually representing an action, offering details not referred to in talk, or enacting projected next actions.

Much of the literature, as can be seen, is addressed to languages other than English, and more importantly, these gesture holds differ from our "halts" in the sense that moving food from plate to mouth is technically not, at least not always, a gesture because it does not function as an utterance or "a component of an utterance" (Kendon, 2004, p. 110), it does not carry meaning, and it is not "co-expressive with synchronous speech" (McNeill, 2005, p. 22). In addition, the gesture holds are followed by retraction, and our halts by resumption. Closest to our notion of halt is Cibulka's (2015) "provisional home position," where the hands are slightly relaxed but held in an intermediate position on the way back to home position. According to the author, such a position is produced to "display a momentary suspension of their pursued line of action and produce a 'just-for-now' stance" (p. 20). Kita (1990) also identifies a momentary "freezing" of movement before and after the stroke (the main phase of a gesture), namely, pre-stroke holds and post-stroke holds. Such holds in co-speech gestures, according to McNeill (2005), "[wait] for a specific linguistic element" in order to "ensure that the synchrony of the stroke and the co-expressive speech is attained" (p. 33). These holds closely resemble what we refer to as "halts" during non-gestural movement.

3. Data and Method

Data come from two large corpora of video recordings of family interaction and adult ESL classroom interaction in the United States, which include (1) 19 hours of dinner conversations with Zoe and her parents recorded at two age periods (35 dinners at the age of 3 and 16 at the age of 8) and (2) 27 hours of classroom interaction collected from a lower-level and an intermediate level adult ESL course. A collection of 37 instances of halt during eating and drinking is built, and not surprisingly, most of these instances (n=35) come from the family dinner corpus. The final collection and analysis of each of its instances is a result of weekly 1-hour data session conducted between the two co-authors over a period of 8 months.

The instances of halt are transcribed and analyzed within a conversation analytic (CA) framework. In addition to using the standard conversation analytic notations, nonverbal conduct is represented in a smaller, italicized font, a dash that connects the verbal and the nonverbal (or nonverbal and silence) represents co-occurrence between the two, and a tilde partial co-occurrence. For example, "nods-yes. sure." indicates that the participant utters "yes. sure." while nodding; "nods~yes, sure" means that the uttering of "yes" starts in the midst of the nodding; absence of the dash or tilde signals that "yes, sure" are produced after the nodding. When necessary, curly brackets are used to demarcate the beginning and ending of the simultaneous occurrence, as in "{nods-yes.} sure," which means the nodding only co-occurs with "yes" but not "sure." Between the two nonverbal activities, a comma indicates sequential (e.g., nods, smiles) and a slash simultaneous (e.g., nods/smiles) happenings. The number sign # indicates the precise moment a screenshot is captured. The notations have been developed to capture visible conduct in as compact and parsimonious a manner as possible.

4. Analysis

Of the 37 cases, 23 halts are done to achieve some sort of completion, and 14 to facilitate repair. Both, as we argue, are exercised to safeguard contiguity. In what follows, we first show 6 cases where halt is used to facilitate different types of completion and then 3 cases of halt used to facilitate repair, all of which feature the prioritizing of speaking over eating and drinking. We end the section with a deviant case, where eating is prioritized.

4.1 Halt to facilitate completion

In this section, we demonstrate how halts facilitate the completion of an ongoing TCU (Sacks, Schegloff & Jefferson, 1974), an ongoing multi-unit turn, or an ongoing sequence. We show two instances of each below.

Immediately prior to the following extract, Zoe was telling her parents about the big machine (x-ray) from the children's TV show "Charlie Brown," and as the segment begins, the family continues eating (line 01) before Dad marvels at Zoe's level of conversation (lines 02-04), thus shifting from the earlier three-party to a two-party interaction between him and Mom, now referring to Zoe in the third person.

Extract 1. Level of conversation

Note that Dad lifts a forkful (line 02) just prior to producing his assessment, halts mid-TCU (line 03; Figure 1.1), or more precisely, mid-word (i.e., after having produced the first syllable con of conversation), and only completes the movement (line 05) after finishing the assessment (also see Erickson, 1992). Perhaps we would agree that stopping mid-word to take a bite would be highly unusual, but what about resuming eating after uttering conversation? For one thing, it would leave one wondering "What about her level of conversation?" It would create suspense, except that creating suspense doesn't appear to be Dad's project at this point-which is to produce an assessment on Zoe's prior telling. The halt allows Dad to preserve the contiguity of his assessment-in-progress. Prioritizing the completion of the movement (i.e., biting, chewing, and swallowing), on the other hand, would have substantially delayed the completion-and breaks the contiguity-of the TCU, assuming that talking with a mouthful would be frowned upon.

In the next instance, it is Zoe who does the halt, and the TCU-in-progress implements, not an assessment, but a question. In this three-party interaction, Zoe has been explaining to her parents the format of the test she took earlier that day that included numbers, letters, bubbles, and columns (data not shown), and in line 01, Mom identifies the test as multiple choice.

Extract 2. How did you know

As can be seen, although Mom's identification is enthusiastically accepted (line 03) and treated as final (line 05) by Zoe, her characterization of the multiple-choice test as the invention of American education is questioned by Dad (line 07) at the same time as Zoe picks up a forkful (line 08). In line 09, rather than await Mom's response to Dad's question, Zoe launches a question of her own (how did you know) as she lifts her forkful towards her mouth. Before taking the bite (line 11; Figure 2.1), however, she halts the movement to produce the clausal object (line 10; Figure 2.2) of the verb know, thereby completing her TCU (and question) in-progress. In the end, it is her question, not Dad's, that gets responded to by both parents (lines 12-14).

Completing the movement without the halt would have similarly delayed the completion of a TCU-in-progress as we observed in the prior case, but more importantly in this case, since Zoe's question skip-connects (Sacks, 1992, p. 349) back to line 01, any break after how did you know (a possibly complete TCU on its own) can be mis-heard as "how did you know it's the invention of the American education?" given its proximity to Mom's response in line 04-the most likely direct object of the verb "know" in this sequential environment. Such mishearing would risk not obtaining any response to her own question then and there, or at all, given the rather different topic potentials afforded by the two questions. What Zoe ends up delivering, enabled by the halt after how did you know, is a single TCU, where it was multiple choice is produced and treated as the direct object of the verb know rather than a new TCU. This is evidenced in both her withholding of pitch drop after how did you know and Mom's answer in line 12.

In other words, sometimes, the TCU-in-progress is in a particularly precarious sequential position, where its on-time completion becomes paramount. And overall, completing a TCU-in-progress without any delay appears to be the "natural" and expected course of action, where any breaks in such progress may be hearable as "for cause" or inviting a mishearing of some sort. And halting the food-to-mouth movement minimizes such distractions and facilitates the undisrupted completions of TCUs-in-progress.

Since any single TCU is a possibly complete turn (see cases above), in this section, we show how gesture halts can also facilitate the completion of multi-unit turns. In the first instance below, Zoe does a halt to extend a response within the same turn. Prior to the extract, Zoe shares that Lily thinks she (Lily herself) is too fat. During a brief repair sequence that clarifies for Dad who Lily is in this three-party interaction, Zoe picks up a piece of bread (the same piece in line 06) with her left hand when Dad asks Who's Lily? (data not shown) as she continues chewing. In line 01, Mom displays her own recognition of who Lilly is after producing an assessment that reframes too fat as a little chubby. This is followed by what is hearable as Dad's commentary directed to Mom as part of the adult conversation on the awful and senseless obsession with body images starting at a young age (lines 02-05).

Extract 3. Alice a bit fat too

In line 07, Mom seeks clarification of whether Lily called herself fat, and in light of Dad's negative commentary just prior, this investigative work is hearable as Mom's attempt to gauge to what extent Lily herself is implicated in this "misconduct." Note that Zoe is quick to confirm that Lily is indeed responsible for this questionable behavior (line 09), producing uhuh as she moves a piece of bread to her mouth (Figure 3.1). Even worse, Lily also calls Alice a bit fat, which Zoe hurries to add after a halt as she moves the bread slightly away from her mouth (line 10-11; Figure 3.2) (see "retraction" in Hoey, 2018). Since uhuh (line 09) is a sufficient response as a second pair-part (that could have closed the sequence) to Mom's yes/no question, not producing that answer extension right then and there could mean not having the opportunity to produce it at all. With the extension, Zoe seems to have made it her business to portray Lily as someone who assesses not only her own but also someone else's body weight, thereby upgrading the "guilty verdict," and it is the halt that enables the timely delivery of what appears to be a crucial addition for Zoe. The sequence ends with Mom's dismissal of Lily's assessments (line 13), and the family returns to eating (data not shown).

The next instance is slightly more complicated with not one, but two, halts, and they are marshalled by Mom during a two-party adult conversation that lays out her opinion of the month January. The segment occurs at the tail end of the dinner, where in a quiet voice, Dad complains about being tired and about riding the subway (line 01), Zoe (who has already left the table) chimes in with her own state of woes (line 04) without obtaining any attention from either parent, and Mom characterizes the trouble as first winter gloom (line 02) and then January being a depressing month (line 05). What this characterization also does is a stepwise topic shift (Sacks, 1992, p. 566) from Dad's personal complaint to season talk, and more specifically, the winter month of January. In other words, it starts something new, where depressing may be seen as a prospective indexical (Goodwin, 1996, p. 384) of some sort that awaits unpacking, as shown in the multi-unit turn that ensues (lines 05-08).

Extract 4. Nobody wants to come back

The first halt occurs in line 06 during the 0.5-second gap (Figure 4.1) where, after having just produced a TCU that harbingers a multi-unit turn, Mom gazes up to Dad as if to secure recipient gaze (Goodwin, 1980). This co-occurrence between the gaze and the halt gives us some basis for claiming that the halt is doing a similar sort of job as the gaze (i.e., securing recipient gaze to proceed with a multiunit turn in this case), and as such, perhaps reinforcing it. Her movement of lifting the spoon continues in line 07 as she completes the next TCU that accounts for her assessment of January being depressing, at which point Dad is still looking down. If line 07 is somewhat unclear given the possible ambiguity of this (which most likely refers to winter gloom), come back in line 08 makes clearer the winter-break-to-spring-semester transition Mom is referencing (given that both parents work for universities with regular semester-based schedules). This redoing coupled with the slight spoon movement towards Dad (line 08) might again remind us of the restart done to secure recipient gaze describe by Goodwin (1980).

Perhaps not by coincidence then, immediately thereafter, Dad throws a quick glance in Mom's direction along with a headshake, and what ensues is Mom's second halt (Figure 4.2) as Dad produces an aligning response (line 09) (both verbally and visibly) that partially repeats, and shows no difficulty in grasping, what Mom just said. What the halt appears to be doing is transform the end of line 08 into a "local pragmatic completion point" (Ford & Thompson, 1996, p. 150) within a larger opinion-building turn. As can be seen, Mom resumes speaking immediately after Dad's turn completion in line 09, with the anaphoric phrasing of that's why recasting prior talk as only a pre- to the upcoming point, where for the first time, Mom explicitly refers, rather than merely alludes, to the spring semester (line 11). What only lurked beneath the surface is now finally out in the open. Thus, one might argue that this second halt, which extends till line 18, is produced in the service of holding the floor for staging the thrust of Mom's opinion, where she explicitly states her dislike for the spring semester and substantiates her assessment with a contrast with the virtues of the fall semester numerated in a three-part list (lines 13, 15, 17-18).

In both cases above then, we observe how halt plays a pivotal role in preserving the contiguity of a multi-unit turn that involves either an extended answer or an extended telling of one's opinion, where the absence of the halts would have cost the answerer or the teller the opportunity to say what they needed to say to complete their projects at hand. Preserving the contiguity of multi-unit turn, as we have shown, also necessarily entails managing contingencies such as holding the floor and securing recipiency.

A final type of completion that halts are recruited to achieve is the completion of an ongoing sequence, e.g., producing a second pair-part of an adjacency pair (Schegloff, 2007). The first instance below is a continuation of Extract (1) earlier, where Dad marveled at Zoe's level of conversation (lines 01-02).

Extract 5. Watching snoopy

What Dad produces in lines 01-02 then is essentially a positive assessment of Zoe's verbal ability, and that makes conditionally relevant a second assessment in the form of agreement or disagreement. In line 06, rather than uttering a simple agreement token, Mom halts her forkful as she slightly lowers the fork (Figures 5.1 and 5.2) to do a telling that offers further evidence of Zoe's soaring language skills (i.e., the latter's use of the "big word" leukemia), thereby "resist[ing] any claim to epistemic authority that may be indexed by the first speaker in 'going first'" (Heritage & Raymond, 2005, p. 15).

Importantly, this timely demonstration of the independence of her assessment (in lieu of a simple agreement token, for example) is rendered possible by the halt in line 06. Any delay can incur an interpretation of impending disagreement, given the preference for agreement after the first assessment (Pomerantz, 1984). Moreover, in the event of exhibiting resistance of any kind, any delay in such expression could arguably weaken the very force of the resistance. There is yet another sense in which we can appreciate the import of such timely delivery. Mom's independent assessment is indirectly conveyed through a telling that displays understanding of Dad's assessment. As Sacks (1992) writes, "'[t]he way that you go about exhibiting your understanding is just to produce another that you intend belongs, given what has just been done [emphasis added]. You can put another item in that is consistent with the sort of thing you figure they're doing" (pp. 112-113). In this case, Mom "figures" that Dad is doing a positive assessment of Zoe's verbal ability, and she produces another item that fits with that assessment (also see line 19). It is imperative that that kind of understanding display is produced in the next position, not later. In short, the halt is employed here to facilitate the production of, not just any responsive action to complete a sequence, but one whose very "identity" hinges on its on-time delivery - its contiguity.

In the next and final segment of this section, which occurs at the beginning of the dinner, it is Zoe who halts her movement to provide a micro-sequentially (Deppermann & Schmidt, 2021) fitted response to Mom's unfolding question. Zoe is sitting at the table eating pasta, Dad is still cooking in the kitchen, and Mom is approaching the table from behind Zoe, having just set up the camera. Zoe's movement to pick up food begins in line 01 as she utters the cut-off is it. While it is unclear where she is heading at this point (and we have no record of what happened earlier since this is the beginning of the recording), Mom in overlap launches a yes/no question (line 02), inquiring whether Neela her pre-school classmate has the same one, referring to the pasta lunch as it becomes somewhat clearer later in line 06.

Extract 6. The same one

In line 04, Zoe responds to the yes/no question with a very slight (barely visible) head nod (Figure 6.1). This is followed by a gaze to Mom with a halt that is likely to facilitate the production of a verbal response (Figure 6.2), except that at the very same time, Mom continues with an increment or different, thereby transforming the yes/no question to an alternative one (line 03). Zoe then extends her halt space, thus readying herself to adjust her response accordingly, which she does in line 05 after a quick thinking-gesture (i.e., glance to side and back to Mom) before resuming. In other words, the halt is initially launched to produce a verbal response to Mom's yes/no question and extended as Zoe continues to monitor Mom's developing first pair-part to prepare a fitted, on-time, response thereafter.

Thus, halt appears to be an essential mechanism for preserving the contiguity of a sequence when a conditionally relevant next turn is due (and its delay or absence would be noticeable), be it an assessment, an understanding display, or a response to an (unfolding) question.

4.2 Halt to facilitate repair

In this section, we show three cases of how halts can also facilitate repair (Schegloff, Sacks, & Jefferson, 1977), where the timely production of repair initiation or repair outcome plays an integral role in implementing a participant's larger interactional project. Extract (7) entails a negotiation of request and offer between Zoe and Dad. In the beginning of the extract, Zoe asks if she could put candles in without specifying the prepositional object--where she wants to place them (line 01).

Extract 7. This my dinner

In line 2, Dad seeks clarification of Zoe's request (put candles in what; line 02), to which Zoe responds by redoing it entirely and picks up a spoonful while producing birthday cake towards the end of this redoing (line 03). This is also when Dad produces the conditional granting (Waring, 2020) in recognitional overlap (Jefferson, 1983). Note that upon hearing this conditional granting, Zoe halts her movement to quickly accept the "deal" with alright (Figure 7.1). But instead of resuming eating thereafter, she lowers her spoon ever so slightly (see retraction in Hoey, 2008), gazing at Dad and pointing at the bowl to initiate a repair, this my dinner? (line 06; Figure 7.2).

Given that acceptance is due after granting, Zoe's halt in line 06 allows her to produce the acceptance without delay, thereby preserving the contiguity of a sequence (see above). More importantly, for our analytic focus of this section, however, we may note that for Zoe, clarifying the terms of the "deal" - what counts as "dinner" — is crucial, and this is done via a repair initiation made possible by the gesture halt (although this would also be a case of halt to complete a multi-unit turn). The repair then must be produced on time, as resuming eating at this point can risk consuming the wrong "dinner," and by extension, fulfilling the wrong terms of the "deal." As shown, eating is only resumed after Dad's confirmation (line 09).

In the next example from an adult ESL classroom, halt is also produced to ensure the on-time delivery of a repair initiation. In the Q&A segment following a student presentation by Rebecca (REB), Misun volunteers to ask a question (line 01) and is given the go-ahead (via gaze) by Bonnie (BON) the teacher (line 02).

Extract 8. To go army

As evidenced in the perturbations (line 04) and silences (line 04, 06), Misun appears to have trouble formulating her question. Her sustained gaze to Bonnie in line 06, then, seems to be seeking some sort of encouragement or confirmation, and indeed, Bonnie treats it as such by nodding and gazing to Misun while reaching for her water bottle.

In line 08, at the same time as Misun finally begins asking a question (is it mandatory), Bonnie's right hand is touching her water bottle (Figure 8.1). The motion of bringing the bottle to the mouth (Figure 8.2), however, is brought to a halt immediately after Bonnie hears the last phrase of Misun's question, go army (Figure 8.3), a problematic item as both the preposition "to" and a definite article "the" are missing. Instead of taking a sip of water, Bonnie halts her movement by lowering the bottle (see retraction in Hoey, 2008) to initiate a repair by way of a metalinguistic term and a partial repeat: preposition. to go:: (line 10). Importantly, given that the Misun's question is directed to Rebecca, as evidenced in her gaze (line 08), drinking right when Misun's turn is coming to an end would likely allow Rebecca, the actual recipient of the question, to proceed with an answer. As such, the halting of Bonnie's drinking movement ensures that her repair initiation is timely - delivered at the transition relevant place, with no gap or intervening talk between the trouble source and the repair initiation, so that the contiguity of the repair sequence is preserved (Schegloff, 2000). Pedagogically, the timely production of repair initiation not only facilitates Misun's self-repair but also shifts the focus of the class from meaning (i.e., asking a peer a question) to form (i.e., grammar work). A delay in repair initiation, we argue, might render both pedagogical goals more difficult to accomplish.

In the final extract of this section, we show how halt is also useful for facilitating the production of repair outcomes, and we examine two such instances: the first by Dad to provide a candidate answer to a word search (lines 05); the second by Zoe to perform an other-correction (line 08). Prior to the segment, Zoe started telling Dad about her lunch at a crepe place earlier in the day that she co-experienced with Mom, and her telling continues as she highlights a noteworthy feature of the place (and the amazing thing is?) (line 01). In this three-party participation framework then, Dad is the ratified addressee of Zoe's telling, with Zoe being a potential co-teller.

Extract 9. Bio-degradable eco-friendly

As Zoe introduces the unique feature the plates are made with sugar cane (line 02), Dad gears up to drink (line 03). Zoe then turns to Dad as she extends her turn with a clause (so they're) while holding a forkful of food (line 04). Before completing her description, however, she cuts herself off and takes an inbreath while slightly lowering her fork (a mini-gesture halt that's not our focus here). Treating Zoe's cut-off as a word search, Dad halts the movement of bringing his glass to his mouth to offer a candidate description (Figure 9.1), an oh-prefaced clausal TCU (oh they're bio-degradable; lines 05-06), which is also hearable as his display of understanding of the telling's upshot. His halt, prompted by what could be a word search, appears to be done to provide Zoe timely assistance - a possible outcome of the search, an interactional task he prioritizes over drinking (which he resumed in line 06).

During Dad's production of bio-degradable, Zoe appears to be resuming eating (line 07). However, immediately upon Dad's completion of bio-degradable, Zoe halts her hand movement (line 08; Figure 9.2) to produce an other-correction, replacing Dad's bio-degradable with eco-friendly (line 09). By formatting her other-correction as a full clausal TCU identical to her not-yet-completed turn in line 04, picking up where she left off, Zoe is also hearably asserting her epistemic right over the telling. Taking a bite at that very juncture would delay her correction and imply an acceptance of Dad's description.

As the three extracts in this section have shown, by prioritizing repair over eating and drinking through halts, participants are able to accomplish interactional projects that are of urgency and importance to them, such as clarifying the terms of a deal (Extract 7), prompting a student's self-correction (Extract 8), as well as providing assistance and asserting epistemic rights over one's experience and telling (Extract 9).

4.4 A deviant case

In the final case below, contrary to what we have observed so far, we see Zoe prioritize eating over speaking. The family has been talking about a recent visit to Dad's childhood friend who has three sons, and Zoe claims liking one of the sons Trevon better (line 01). In response to Mom's account solicitation (lines 02 & 05) (Bolden & Robinson, 2011), Zoe claims knowing "why" (line 06) without actually providing the account while picking up a spoonful.

Extract 10. Like Trevon better

In line 07, Mom pursues the account at the same time as Zoe opens her mouth for the spoonful. Upon hearing Mom's why, however, Zoe does not halt her movement but continues on with her eating trajectory (line 09; Figure 10). It is only after the biting and some chewing that she provides the answer sought (lines 10 & 12). In this particular case, Zoe is the only one who is able to answer the question, as she emphatically claims in line 06 with a stress on I. And having secured a multi-unit turn with the pre- in line 06, she is not in any danger of losing the opportunity to produce the answer to Mom's question. In other words, contiguity is not at stake to warrant any gesture halt.

5. Discussion and Conclusion

In this paper, we have shown how the halt of eating and drinking is an "essential worker" in the preservation of contiguity in interaction, i.e., ensuring the timely production of what is due next, the delaying of which may either lose its relevance or weaken its impact. By facilitating the timely completion of a TCU/multi-unit turn/sequence-in progress, halt manages a variety of contingencies such as minimizing potential mis-hearings, ensuring extended telling or answering, securing recipiency, and avoiding any noticeable absence of a conditionally relevant assessment, understanding display, or response to a question. By facilitating the on-time production of repair initiations and repair outcomes, halt also enables the participants to sidestep any mishaps that can compromise their myriad interactional projects-in-progress, be it clarifying the terms of a deal or asserting one's epistemic primacy. A strong piece of evidence that halt is indeed exercised to preserve contiguity is its absence when contiguity is not at stake, as made evident in our deviant case. As can be seen, compared to the "prestroke holds" (Kita, 1990) that are designed for the sole purpose of synchronizing talk and gesture (McNeill, 2005), our non-gestural halts addressee a much wider range of interactional concerns.

Our findings contribute to the broader literature on multiactivity and materiality. As shown throughout our instances, the manipulation of cutlery is part and parcel of managing the multiactivity of eating/drinking and speaking. In particular, the initial movement of picking up food or drink can co-occur with (e.g., Extract 1) or precede (e.g., Extract 8) -- during another's talk -- the beginning of one's own TCU. Halting the movement of cutlery or drinking vessel, on the other hand, is found mid-TCU (e.g., Extract 2), at the transition space (e.g., Extract 1), or during another's talk (e.g., Extract 6). Such halts are produced to allow for speaking (e.g., Extract 5), project further talk (e.g., Extract 4), or signal incipient speakership (Extract 6). Finally, resumption of the cutlery/drinking vessel-to-mouth movement (that culminates in the taking of a bite or a sip) follows, thereby double-marks, the completion of talk. As shown, the co-participants often start speaking after such resumption (e.g., Extract 4) although in some cases, they start at the same time as the resumption -- immediately after the other's verbal completion (e.g., Extract 9). Not surprisingly, our halt cases all manifest the "exclusive order" of putting one activity on hold in favor of the other (Mondada, 2014), and more specifically, putting eating on hold in favor of speaking in the interest of preserving contiguity.

Unlike the cases in Erickson (1992), our participants' reaching for food is not always synchronized with others' speech. At the same time, our findings, as described above, also offer greater specifications of Erickson's observation regarding the coordination of "food to mouth" sequences with various boundaries in talk (e.g., syntactic and prosodic junctures, intra-clause and inter-clause boundaries). Similar to those in Hoey's (2018) cases who resort to acceleration or retraction, our participants also adjust their drinking (and eating in our case) trajectories to accommodate speaking, except that such adjustment entails a somewhat distinct suspension of the movement, which we call "halt," although some of our halt cases do involve a very slight retraction (e.g., Extracts 3, 5, 7, and 8). More importantly, we hope to have extended Hoey's work by illuminating how such adjustments are deployed to accomplish interactionally.

Finally, we hope to have extended the existing scholarship on the "freezing" of movement by (1) broadening its core linguistic and contextual reach beyond what has been mostly European languages and sign languages and (2) establishing halt as a distinct practice that exhibits 'micro-sequentiality' (Deppermann & Schmidt, 2021). In particular, while the contiguity requirement of sequence organization drives the production of halts, it is in safeguarding the on-time completion of a TCU, a multi-unit turn, a conditionally relevant next item and the on-time completion of repair within the repair opportunity space that the very integrity of sequence organization is upheld. In that sense then, our analysis simply provides yet another exhibit of the interdependence between multimodality and sequence organization.

Acknowledgements

An earlier version of this paper was presented by Carol at the 2022 American Association for Applied Linguistics (AAAL) Conference in Pittsburg, USA, where we initially grappled with the distinction between "hold" and "halt." We thank the audience for a productive discussion on that matter. The final manuscript has benefited a great deal from the detailed and incisive comments from the two anonymous reviewers for Social Interaction. Finally, deepest gratitude to our editor Kristian Mortensen for engaging with our work in a way that any writer would dream of.