Several prevailing conceptions of syntax have placed it entirely within the mental realm, arguing that it is an autonomous system of abstract rules. This definition has excluded the bodily production of speaking from the domain of syntactic analysis and fails to adequately address the role of syntax in embodied interaction. The so-called written language bias in linguistic theories (Linell, 2005) has furthermore resulted in the omission of not only the body, but the situated emergence of syntax together with the body in everyday talk. Previous studies that investigate the connection between the body and language have largely treated language as individual and non-interactionally situated (an early example being Lakoff and Johnson, 1980).

These studies assume that grammar is a self-contained, mental pattern highly abstracted from existing or imaginary sentences and independent of actual use (e.g. Evans, 2006; Radford et al., 2009). As a result, any given language’s grammar is treated as a consistent, predefined structure, and deviations as irrelevant chaos or poor performance (e.g. Chomsky, 1965; Newmeyer, 2003). In contrast, studies of situated language use have not only found new and highly organized practices for managing syntax in everyday activities (and this paper reports another such practice), they have also presented cross-linguistic evidence that grammar, including syntax, is a temporally unfolding resource for interaction (e.g. Hopper, 1987, 2011; Hakulinen, 2001; Auer, 2009; Linell, 2009; Fox and Thompson, 2010; Pekarek Doehler et al., 2015). In resituating syntax in its natural ecology, which, after all, is language’s most common site of occurrence, and its evolutionary niche (Bögels et al., 2020), these paradigms have been able to show how language interacts with other semiotic resources, such as prosody, gaze, gesture, and the moving body (e.g. C. Goodwin, 2018; Hayashi, 2005; Keevallik, 2013b, Keevallik, 2018; Mondada, 2014). This paper continues the quest of resituating syntax in the living world and living body, by focusing on progressivity within clauses.

Clauses are one of the central resources that syntax structures and they also constitute core units for turn taking in interaction in several languages (Helasvuo, 2001; Thompson and Couper-Kuhlen, 2005). Clauses are neither pre-established in terms of an exclusive set of fully specified patterns, nor are they necessarily fully pre-arranged in the minds of speakers before an utterance is “sent”. Instead, clauses are emergent over the course of time, dependent on other participants actions and creating possible moments of turn-completion for next speakers to be able to take their turn (Schegloff, 1996). Clauses have also been shown to emerge in close coordination with bodily action. Goodwin (1979) provided a famous example in English that demonstrated how an utterance need not be produced as a single, independent, coherent proposition, but may emerge incrementally in reaction to gaze by different participants. In a similar vein, Iwasaki (2009, 2011) has shown how clausal turns in Japanese emerge as a result of dynamic interactive processes, which include “interactive turn spaces”–places where the speaker of the unit-in-progress invites the interlocutors to co-participate in the building of the action.

Alternatively, a speaker may abandon a turn-in-progress as soon as the action has been treated as complete through the recipient’s bodily visual response (Ford et al., 2012, p. 206; see also Li, 2016). The body can even complete verbal clauses, as if filling in the silent slots left after incomplete syntax (Keevallik, 2013b). Verbal syntax can be discontinued for an embodied demonstration after verbs, copulas and quotatives, but also adverbial phrases, adjectives, articles and subjects (Keevallik, 2015). These embodied displays effectively take the position of lexical units; not only do language and gesture complement each other when making meaning in so-called composite utterances (Enfield, 2009), but an embodied demonstration can actually occupy a grammatical and temporal slot within the emerging syntax (Keevallik, 2013a). It has, for example, been shown that a speaker can complete a grammatically incomplete structure with an explanatory gesture, such as in word searches (Hayashi, 2003) and in second language conversation (Olsher, 2004; Mori and Hayashi, 2006). Slots for embodied displays can also be projected by specific phrases, such as und ich so/und er so “and I’m like/and he’s like” in German (Golato, 2000) and it’s like in English (Fox and Robles, 2010). The slots filled by bodily displays even function similarly to verbal turn components (Ford et al., 2012; Keevallik, 2018), suggesting a strong emic basis for incorporating these bodily options in a descriptive grammar. In short, we have begun to see how various aspects of embodied behavior participate in and steer emergent grammar.

Given that syntax is dependent upon a living body in order to be expressed (from airflow in speaking, to motions for typing, signing, or machine-assisted talk), it is remarkable that so little work has addressed the connection between the situated living body and syntax in interaction. This paper will target that gap, focusing on scenarios where bodily involvements are made relevant for syntax production, and where the connection between syntax and body becomes a member’s resource for a particular action through the phenomenon of strain.

Progressivity is a central concern for participants, both regarding the advancement of interpersonal interactional sequences as well as internal components of individual speaker turns (Schegloff, 2007, pp. 14–15). Nevertheless, disruptions ¹ to syntactic progressivity occur in many different scenarios in interaction; parenthetical clauses, word searches, self-repair, certain question formats, and collaborative completions all exploit the normatively through-produced nature of syntax, and are dealt with in systematic ways by the interacting parties. For example, parenthetical clauses can be inserted into clauses that were already in progress and projected to be complete, which results in temporarily halting, then resuming, the parent clause’s syntax (Mazeland, 2007). This “enables a speaker to reconcile the potentially contradictory requirements that the linearity of speech production poses to the speaker’s orientation to recipient design” (Mazeland, 2007, p. 1816). Word searches, a type of self-repair, constitute another instance of compromising progressivity, as they are also initiated by a disruption in the utterance, resulting in a delay in the middle of an ongoing turn-constructional unit (TCU) (M. H. Goodwin and Goodwin, 1986; Hayashi and Yoon, 2006). In this paper, we are likewise focusing on structures that involve the process of suspending a clause that was projected to be complete, rather than just any word combinations. While mid-TCU silences have mostly been connected to cognitive trouble, such as in word searches, we will argue that they can also be related to bodily concerns.

Mostly, disrupted syntactic structures (including those lacking embodied completion) are treated as inadequate for achieving a complete turn-at-talk. Where speakers do leave a phrase incomplete, it is marked; it does something in the interaction. Leaving an utterance entirely unfinished can be a way to leave something delicate unspoken, inviting co-participants either to produce a completion of the item (such as in bad news deliveries, Lerner, 2003) or to align by treating the unfinished phrase as adequate for progressing the sequence (as discussed in Chevalier and Clift, 2008). Disrupted syntax is one way to designedly invite collaborative completion or “recipient intervention” (Lerner, 2013, p. 125). One format, “fill-in-the-blank” questions (Persson, 2017), is often used in institutional environments such as teaching (Koshik, 2002). Collaborative completions may take advantage of compound TCUs underway to complete the syntax and thereby demonstrate affiliation, agency or epistemic rights, through co-authorship of the statement: the collaborator in this case “co-authorize[s] what is being done” (Lerner, 2003, 2013, p. 126). However, the practice described in this paper involves a more abrupt suspension of syntax, often through cut-offs, which is more commonly used for initiating self-repair (Fox et al., 2009). Furthermore, in self-repair, speakers often reframe the repair solution in a way that backtracks to a prior noun, verb, or prepositional phrase (Schegloff, 1987; Fox et al., 2009), thus restarting the syntax. This provides recipients with the needed resources to determine whether the talk following a disruption is, indeed, a repair of the prior talk, or a continuation of it (Sidnell, 2006a). In moments of repair, by reframing the post-disruption talk to re-start the syntax, speakers can situate the trouble solution with respect to the original phrase.

As we will show below, such reframing is rare in our suspensions; they often begin where they left off, so to speak, unless competing talk forces them to re-attempt a suspension (see Interactional Treatment of the Practice as Coherent). With these considerations in mind, we can define this paper’s phenomenon more specifically: a syntactic suspension is a moment when an initiated clause (projected to have complete clausal components) is temporarily halted, resulting in a silence, before being resumed, at least briefly, along one of the potential clausal continuations that was projectable before the suspension began (see Methods for further details). We will show how the practice of syntactic suspensions connects the progressivity of a clause to a developing embodied activity. The study will thereby further inform our understandings of how participants orient to both syntax and the body as reflexively constituted social phenomena.

People use their bodies for both physical movements and speaking. Speech is produced through the same anatomical structures that ensure the maintenance of life through breathing: the lungs, the larynx, and the vocal tract. Muscles of the breathing and vocal apparatus control whether air can pass through the glottis. With respect to breath tasks relevant to this paper, these muscles must manage breath holds, glottal closures, as well as participate in maintaining sufficient air pressure for speech (Lieberman and Blumstein, 1988; Hixon et al., 2018; Fuchs and Rochet-Capellan, 2020). Strenuous activities alter speech breathing and speech production, as the body’s need for oxygen increases, resulting in additional physical effort should one want to speak, as well as perceivable, effort in the voice (Sandage et al., 2013; Trouvain and Truong, 2015). Even with strain, however, speakers are often able to constrain breathing to occur at syntactic boundaries (Trouvain and Truong, 2015; Hixon et al., 2018), which maintains the coherency of the emerging syntax.

Many of the same muscles used for breathing are implicated in establishing certain forms of bodily stamina, tension, and steadiness. Physiologically the human body has more stamina when the torso and breathing muscles are engaged, including laryngeal muscles that close the glottis (Hodges et al., 2005), as is sometimes the case during breath holds (Massery et al., 2013) or controlled/forced exhalation (Ikeda et al., 2009; O’Connell et al., 2016). When these muscles or breathing techniques are used, the body can undertake more force or exert strength longer (Welch and Tschampl, 2012). Even untrained people spontaneously hold and control their breath when lifting objects (Mateika and Gordon, 2000; Lamberg et al., 2003; Hagins and Lamberg, 2006). Since breath and muscle tension are necessary for language production, when these systems are bound up in other activities, a speaker may face trouble in producing normal speech. As we will show below, however, such difficulties may actually provide affordances for displaying strain.

As the above prior research demonstrates, incomplete syntax is taken as a sign of some trouble on the part of the speaker (from finding an appropriate word to having sufficient breath), and co-participants either withhold talk until progressivity is resumed or provide assistance to progress past the trouble. Sometimes the disruption arises due to communicative deficits. In the utterances of people with aphasia and other communicative disorders, turns are characterized by multiple silences that punctuate the TCU at inapposite moments (Laakso and Klippi, 1999). Stammering creates disruptions on the phonemic, syllabic or word level (R. Wilkinson and Morris, 2020), while anomia and agrammatism may occasion perturbations due to inability to produce the next relevant word, and dysarthria may lead to frequent repair (Auer et al., 2020). Progressivity halts in these interactions often make the turns of persons with communicative deficits vulnerable to completions by other interactants. Whether such collaborative completions are an intrusion or a resource is unclear (Aaltonen and Laakso, 2011; Norén et al., 2013). “Intrusions” can permit speakers to progress the interaction more smoothly (C. Goodwin, 1995; Perkins, 2003) with the result that “the attention to the aphasia is minimized” (Oelschlaeger and Damico, 1998; Bloch and Beeke, 2008).

The syntactic suspensions in the current paper are attributable to bodily strain and the body being given prominence and precedence over speech (i.e., the bodily activity is progressed even if speech is delayed, much like in some cases of multiactivity, Haddington et al., 2014). The focus of study here is not bodily deficits or disability-related speech disruption and, to the contrary, we will show that participants go to some work to demonstrate (partially maintained) control over the timing of the suspension and resumption of their talk (What’s in a suspension?, Interactional Treatment of the Practice as Coherent).

In short, research on atypical interaction demonstrates how incomplete syntax is treated as accountable and possibly a sign of real difficulty to continue. Word searches in all kinds of contexts demonstrate the same occurrence on a more temporary basis. In the following, we analyze when and how participants employ the syntactic suspensions as a display of temporary incapacity to maintain simultaneous speech and body movement. This paper will thus contribute to our understanding of how the body and syntax are co-organized, as well as to how bodily strain can be displayed in everyday interaction.

In this first section, we describe the features of suspended syntax. The practice is characterized by a suspension occurring in the middle of an emergent syntactic clause (occasionally in the middle of a word), which is subsequently resumed. The suspension takes the form of a silence, although strain sounds or outbreaths can co-occur with the silence (see the uvular trill in Extract 1, further analysis in What’s in a suspension?). The suspension is timed with a moment wherein the body is affected by strain, often the peak of a strenuous movement, but sometimes the out-of-breath “after-effects” of strain. The continued clause starts where it was suspended, without reframing (as often done in repair, see Extract 6, Extracts 10, 11 for deviant cases).

The following (Extracts 2–5) each show a rock climber suspending their talk at a moment where they perform a strenuous motion. They are of course using their bodies throughout the encounters, and they are each already climbing by the time the extracts begin, but the suspensions occur at some difficult motion. The suspensions occur in the middle of a word (Extract 2, line 5, see Figure 2), a verb phrase (Extract 3, line 2, Extract 4, line 3) and an adverbial phrase (Extract 5, line 5). At each of these points, the TCU is grammatically incomplete, and transition to another speaker is not yet relevant.

As the climbers engage their bodies in additional exertion (especially the abdomen, including the diaphragm, which is associated with the function of breath holding, Parkes, 2006), they suspend their talk. The exertions above include lifting a foot/leg (Extract 2, line 5; Extract 4 line 2–3), raising a hand while lifting the body (Extract 3, line 2; Extract 5, line 3–5), and catching oneself during an unexpected slip (Extract 4, line 3). The suspensions occur in the middle of a projectable clause, at moments of “maximum grammatical control” (Schegloff, 1996, p.93). For example, beginning with Extract 2, when Pat suspends his speech at “to hold myh” (Extract 2 line 5), he has projected either a noun phrase (e.g. my body) or a compound word (what he ends up producing, myself). This is not a point of possible completion for a TCU (it is, currently, a “fragment,” Selting, 2001), and so not a place for the co-participants to take a turn. Pat completes the clause starting the moment his foot gets in position underneath him, which coincides with the moment his muscles can relax, relatively speaking, since he is more securely supporting his weight.

In contrast, when any suspension occurs after a point of possible completion, a co-participant does take a turn. For example, in Extract 4, after two suspensions from Quin that occur in the middle of clauses, the third silence (line 3, 0.2) occurs at a possible completion point of the clause: I just wanna do it. Adam treats this as a potential transition relevance place, and takes a turn (line 4), overlapping Quin’s increment (line 3). Thus, even though Quin has just slipped, is currently in the process of stabilizing his body, and may be temporarily unable to continue speech (all of which Adam can see), Adam still treats this particular silence (but not the previous ones) as a transition relevance place. This suggests that the position of the suspension in the middle of a clause, at a position that is not a possible completion point (Schegloff, 1996), is critical to the achievement of the suspension practice. In suspending at moments where the syntax is incomplete, the speaker takes advantage of the normal projectability of syntactic completion. When a clause is incomplete, all of the co-present climbers in the above examples treat the turn “space” as unavailable, as in the middle of the climber’s TCU and not an opportunity to take a turn. They do this by withholding talk until a possible completion point. Participants can hold the strain display over significant temporal delays (e.g., 1.7 s, Extract 2 line 5), without co-participants treating the silence as abandonment or a lapse (Hoey, 2020), due to the availability of the bodily preoccupation as an account.

Bodily control is often, though not always, regained at the moment when the syntax is resumed, such as in Extract 5 (line 5): Pat is here trying to balance and find the next hand hold. He displays strained voice while standing up and holding on with only one hand, but once he gets his legs fully extended and stops reaching with his hand, he has better balance and weight distribution. At this point, the strained voice also stops, and the clause is resumed. In Extract 3, where the strained voice is instead added after the syntax is resumed (line 2), the climber has not yet quite reached a moment of rest, although he can anticipate reaching it when he gazes up shortly before resuming the clause (line 2). The climber’s utterance describes the hardest part of the climb as it is happening, so the remainder of the clause is squeezed out in a strained voice. Continuing the clause, through ongoing strain, displays a barely maintained ability to utter speech. These complementary examples show how suspending syntax permits the straining body to display that it has reached a moment when speech is temporarily unsustainable with bodily activity, but also creates an opportunity, by holding the turn, to demonstrate the prospective resumption of control.

That this is a display, not just a physiological byproduct, is evident in that the speakers are capable of timing their speech to occur in moments of strain. For example, in Extract 2, Pat produces a complete clause without a suspension (line 3). This co-occurs with hanging, rather than moving; Pat is relatively relaxed and unstrained in this position. Climbers are not forced to move at any given point, nor are they required to talk while climbing (as above) or demonstrating a motion (as with opera rehearsals, below). Yet Pat continues to speak (line 5) when initiating and undergoing a movement that causes bodily strain. By designing or permitting talk to co-occur with motion and strain, the participants create the conditions wherein a syntactic suspension can function. In other words, timing talk to co-occur with motion and strain provides a slot in which a display of that strain can be done with suspended talk, as the body provides a visible account for suspending syntax. Speech can be prioritized even over physiological needs, such as full ventilation (Hixon et al., 2018), and the syntactic suspensions take advantage of the ability to switch between priorities, live. The body’s motion is concurrent with the entire utterance, however in the silent moment, the body is rendered as the most salient part of the display—by removing language production, the body becomes the main focus of attention. The suspensions are timed to strategic moments that highlight relevant issues for the speakers, such as a challenging body position or move.

Suspensions do not only occur in canonically bodily intensive activities like exercise. In Extract 6, at an opera rehearsal, the body is likewise made salient at the moment when a minor strain becomes too great. This is an example of Swedish opera performers and their director trying out initial ideas on how to embody a scene. Since the baritone (BAR) will be mourning over the soprano as she dies, the director (DIR) has suggested they try it with the baritone seated on the floor, and the soprano (SOP) lying in his lap. This creates a difficult position for singing. In opera performance, effort should, of course, not be revealed in the voice, so it is relevant for participants to note and remove strain in rehearsal. In lines 1-2, the baritone launches talk about a scene and characterizes it in a copula clause de e såhär handfast “it’s hands-on like this”. He then begins to experiment on line 4, launching a proposal of how he might move här måste han liksom bara “here he just has to,” before trying to lift the soprano into his arms. The lift (lines 5–11) takes several efforts to accomplish (see Figure 3), across which the baritone suspends syntax.

The first suspension occurs after the adverbial bara “just” when the baritone is lifting the soprano. In the silence, the director co-animates the baritone’s proposal by voicing the musical lines that he should perform at this point (line 5). However, as an instance of our focus phenomenon, in line 7, the baritone produces a continuation and completion of his launched syntactic structure: här måste han liksom bara ta e t- tag “here he just has to take hold/grip” (lines 4–10). This emergent completion retroactively characterizes the incomplete syntax and his silence (line 5 and line 8) as part of a suspension borne of strain, rather than the opening of an embodied depiction (which “bara” can project, Eriksson, 1995). The abandonment of ta- (line 7) coincides with a slightly higher lift by the baritone, and together with the following pause (line 8), this achieves yet another display of strain, albeit in the form of repair (particulars of how our phenomenon unfolds with repair will be discussed in Interactional treatment of the practice as coherent below.)

The baritone does further work to achieve the suspension as strain-motivated. His prosody from här måste onwards is marked by strained voice, as with the climbing examples, demonstrating the tightness of his muscles. The baritone produces this strained voice even when anticipating, and not yet doing, the lift of the soprano (line 4). Later, when stating his uncertainty about how to do the action (line 11), there is no strained voice, even though he is still lifting the soprano up. The strain is focused at the moment of initiating the action, helping to project for the soprano that the lift is beginning, and for both Soprano and the director that the lift is effortful—something that may be highly problematic in the product of operatic singing later.

We can thus see across activities that suspended syntax renders salient the current strain in the speaker’s body, as a social action. The practice displays a balance between ostensibly being overwhelmed by bodily strain (temporarily suspending talk) and control (being able to continue talking). The periods of silence coincide with a successful achievement of a strenuous move. Syntax is interrupted at moments of “maximum grammatical control,” where the TCU is accountably incomplete through the launched-but-incomplete-clause, which helps hold the turn for the current speaker so they may resume the syntax after the silence. In the next section, we compare the different practices used to display strain at moments where syntax is interrupted.

Once syntax is suspended, speakers have several options for the unfolding turn, including producing silence, hearable outbreaths, or grunts. Silence is coordinated with strain peaks and bodily motion (see Interactional treatment of the practice as coherent below for deviant cases). Extract 7 below demonstrates the silence format, with the spectrogram (see Figure 4) showing the lack of sound emerging in the suspension. In this example, Quin suspends syntax in the midst of complaining about the beginning of the climb. The suspension emerges just before he reaches (“LH over”) for another hold.

As Quin selects and moves to a better position, he temporarily takes up more weight in his other muscles, causing additional strain. This occurs just as he resumes syntax, resulting in the word “not” being produced with strong emphasis and a rise in pitch. These prosodic changes have a plausible physiological connection: the extra pressure in the vocal tract may arise from muscular adjustments involved in increasing stability in the torso (Hodges et al., 2005; Massery et al., 2013). The enaction of this muscular effort creates that bodily event of stability, because the muscles must assume the stability-providing shape in order to produce the prosody, even if the stability itself is not strictly required. In other words, whether the display arises from airflow requirements, or airflow is suspended in order to do the display, the physiology and social action are inextricably bound together.

Another way to enact ongoing bodily effort is to preface or intersperse the suspension with breathiness or outbreaths, as if “catching one’s breath” (a feature also found in experimental instances: Trouvain and Truong, 2015), likely reflecting the body’s increased need of oxygen due to the strenuous activity (Hixon et al., 2018). In Extract 8, the Swedish-speaking climber (CLI) has just completed a climb and is looking down from the top of the boulder. He is breathing heavily (line 2, 6), and when he appears to have enough breath to talk again, he begins to assess his climbing performance with a beginning of a routinized assessment format vad “how”. However, before continuing the syntax, he takes another breathing break (middle of line 8), and still does not make it to a grammatical completion. The next word is creatively assembled, rather than a conventional continuation (oform “nonshaped/out of shape”) but as an evaluative item it pragmatically continues the projection from the format-initial item.

The outbreath in line 8 manages the body in a different way from silence: instead of appealing to the embodiment of breath holding (and its relevancy for torso stability in a difficult motion), the heavy breathing displays that the speaker enacts having experienced strain and currently not having enough air to complete the projected phrase. As shown above, this may be a way to accomplish a display of strain, via suspension of syntax, after the strain event has concluded (see also Extract 2). The climber is not actively undergoing the difficult motion at the time of speaking; he has just completed the climb. Heavy breathing may be physiologically necessary after strain, but it can also be deployed at specific moments to accomplish a syntactic suspension; the climber chooses to describe his climb now, rather than wait for his breathing to ostensibly be normal. In this way, both glottal closures and heavy breathing can occur during the intra-TCU silence, and tie the display to slightly different bodily demands. In a similar manner, Pehkonen (2020) has argued that by using the semi-conventionalized Finnish token huh huh participants externalize their physiological state of “being out of breath” and take a stance towards the triggering event. Suspended syntax seems to be another practice of stance-taking; the strain displayed by the halt in speech production orients to the prior event(s) as strenuous, and as having various qualities (tough, involving excess strain, etc.).

There is yet a third option in formatting suspensions that are associated with strain: producing a strain grunt. Grunts were briefly described by Goffman (1978, p. 803), as sounds that serve to warn others that “nothing else can claim our concern,” similar to how we have described the way syntactic suspensions bring the body to prominence. Grunting in our data consisted of a short burst of guttural vocalizing, usually with a glottal onset and low open or mid vowel. Most grunts in the climbing data corpus (365 out of the 368 found) did not interrupt syntax, either occurring as stand-alone utterances, or preceding or ending an utterance that had a complete clause. The few grunts that interrupted syntax resulted in the abandonment of the clause, rather than its temporary suspension, as the following extract shows. Below, Pat is trying to find a safety bolt in a strenuous position. At first, he uses syntactic suspension (line 2), but after a long pause and a grunt, the clause remains unfinished and instead he launches another clause.

The clause that Pat started suddenly becomes irrelevant, as he discovers the safety bolt just before the silence ends. The grunt (line 2, “enh”) occurs as Pat repositions his hand to have proper leverage to connect his rope to the safety bolt. In other words, once it becomes pragmatically unnecessary to complete the clause, the suspension is interrupted, in this case by a grunt. The grunt reorients to the physical effort involved in grasping the hold and using the bolt, and may act as an upgrade of further strain that occurs while a suspension is already ongoing. This suggests that grunts may allow exit from a suspension that can no longer be achieved coherently.

In contrast, similar strain sounds may provide reentry into the completion of a suspended phrase. Reconsider Extract 1 (line 3, transcribed as “R”), the word much is prefaced by a uvular sound, and the initial m is lengthened and infused with strain, thus delaying syntactic progressivity. The seeping of air through the strained vocal tract may indeed index inability to continue but, in Extract 1 at least, the syntactic structure nevertheless emerges as complete. In fact, the strain sound may preempt co-participation by getting vocalization on record, and then merges the sound into the opening of the syntactic resumption. A grunt produced as a standalone token, as Extract 9 above shows, may provide the opposite affordance. As in all other instances of suspended syntax, this carefully balances displaying physical preoccupation and sufficient control of the vocal apparatus to intermittently produce speech.

Silence, outbreath and strain sounds are thus treated as different components of the syntactic suspension practice. While grunts and breathiness can also achieve strain on their own, syntactic suspensions uniquely allow the speaker to display re-established bodily control, through the (smooth) resumption of syntax.

We argue that syntactic suspensions are not merely by-products of bodily demands, even if they are interwoven with such demands. As evidence that they are a member’s practice (Garfinkel, 1967; Sacks, 1992)–that is, recognizable and reproduceable for pragmatic action–we now demonstrate below that speakers orient to, and work to maintain, the coherency of the suspension as a practice. Syntactic suspensions are not just any emergently formed clause with a discontinuation. The suspension occurs at specific moments and alongside systematic practices for excluding co-participation in the clause. This section will analyze how the strained participants actively restrict co-participation in the clause, to allow for achievement of the clause as if temporarily suspended due to strain.

Let us first return to Extract 6 (reproduced below as Extract 10), wherein the baritone was demonstrating how he might lift the soprano into his arms. In this extract, we can see the inherent risk in using suspended syntax as an effort display; especially in some syntactic constructions it is vulnerable to being treated as a syntactic preface to a bodily completion (line 4) (Keevallik, 2018). The syntactic structure that has been begun is här måste han liksom bara “it’s hands-on like this and here he just has to like and comes to a halt after bara “just”. This could introduce a bodily completion (the lift, line 5 onwards), and the director appears to treat it as such by providing the relevant sung line that would accompany that movement. However, the baritone subsequently extends the clause (line 7), eventually resolving the syntax as an instance of suspension arising from strain.

During the opera rehearsal, bara or its shortened form ba (Eriksson, 1995) is frequently used to introduce depictions–embodied iconic representations of distal “scenes” (Clark, 2016), also known as for instance animations (Cantarutti, 2020), reenactments (Sidnell, 2006b) and reported speech (Holt, 2007). It is therefore possible to treat the syntax as completed by the bodily movements of the baritone (see Keevallik, 2018); his embodied depiction can act as an alternative to the complement of the projected clause. The director treats the ensuing actions in this way, by positioning herself as a co-animator of the embodied depiction and voicing the musical lines that he should perform at this point (line 6). Furthermore, the strain involved in the lift is visually available, as the baritone has not yet finished successfully lifting the soprano and appears to be repeatedly readjusting his grip. Co-animating at this point supports the baritone’s depiction by providing libretto material he may not be able to produce. However, the baritone does not orient to the director’s co-depiction, overlapping it with an utterance voiced as himself, the performer, rather than as his character: he describes his attempts to get a grip on the soprano in order to lift her (line 7–10), in synchrony with his physical attempts to do so. The director, in response, likewise switches from a depiction of the baritone’s character to commentary as herself, a director, in the current action (line 9, a continuer). The baritone thus bypasses the director’s contribution in the depiction, reorienting them both to the ongoing physical strain in his performing body. In retrospect, this emerges as a syntactic suspension of the structure liksom bara ta tag “has to like take hold” whereby the director’s singing ends up being treated as a side activity.

Suspensions are in general vulnerable to co-participation, since, if the pause is long enough, they can encourage co-participants to treat the ongoing clause as either abandoned or in need of assistance to be completed (M. H. Goodwin and Goodwin, 1986). For example, the director, above, supplies the baritone’s sung lines in the absence of the baritone providing them himself. However, in the following example from rock climbing, when such assistance is offered, the original speaker rejects the contribution, and reattempts the suspension in the clear of co-participation. Below, Pat has just abandoned doing a “roof” climb that was too hard for him that day. As his feet touch the ground, he appears to assess his shaking arms (Lookh at thath, line 1) as notable, before suggesting an alternate next climb. His speech is marked by breathy phonation (see also Extract 8).

When Pat suspends his syntax after Let’s do sohme (line 3), Eric at first withholds co-participation, possibly until Pat regains his balance (line 4). However, after a lengthy silence (line 4), during which Pat gazes at Eric (lines 3-6, see Figure 5A), Eric supplies a possible option, not a roof, which pragmatically, although not grammatically, completes the TCU (line 5). This co-participation ends up being in overlap with Pat’s next installment of the ongoing clause (an “eight” or “nine” being a type of climb, specifically a degree of difficulty, much as a “roof” is a type of climb, though not an official difficulty grade) (line 6). Instead of agreeing with Eric’s suggestion (e.g., by uttering “yeah,” or “exactly”), Pat restarts his syntactic completion (lines 8–10), including an insertion (S. Wilkinson and Weatherall, 2011) of nines. Insertions between the first and second half of the suspended clause are unusual for the collection, but co-participation is also rare. Pat also withdraws his gaze from Eric here (see Figure 6B), reducing the relevance of a collaborative completion. Furthermore, Pat not only re-starts his attempt to complete the prior suspension, but also initiates a second suspension (between lines 12-14, see Figure 5C). In this way, Pat gets a syntactic suspension on the record that is neither influenced by a co-participant, nor modified by any sign of self-repair. Pat also adds extra practices for displaying strain that were less present in the first attempt: there is more breathiness (lines 10, 12, 14), and also a visual display of his tired body (line 12). In this case, the suspension does not coincide with a strenuous move itself but rather with a display of physical exhaustion from prior strain that ostensibly prevents the speaker from producing continuous syntax (see also Extract 8). Thus, in contrast to suspensions during strain that featured breath holds and strained voice, this one involves heavy breathing. At the same time this instance again evidences the careful balance between being too exhausted to do a roof but still sufficiently capable of uttering snippets of syntax. Exhaustion is here designated by the breathy voice and intermittent panting and speaking.

Both Extracts 10, 11 exhibit self-repair work, ta e- ta- ta:g “take hold/grab” and eights- nines and eights respectively. Most instances of our suspension collection do not involve any insertion or reformulation after suspension, but instead directly resume the clause where it was suspended. Notably, both of these extracts with self-repair modifications occur in moments with overlap. In Extract 10, the director first receipts the baritone’s utterance as prefacing a demonstration syntactically, and so takes it that the turn space is open vocally. In Extract 11, Pat meets Eric’s gaze (lines 3–6) during the silence, which can be a strategy for inviting a co-participant into a word search (M. H. Goodwin and Goodwin, 1986). Eric takes this invitation and supplies a potential completion. In both cases, redoing a suspension gets it on record as a suspension, rather than as some other action with co-participation (e.g., demonstration, word search or other inability to complete sentence).

The coherency of the syntactic suspension as a social action rests on being able to maintain silence without intervention and resuming the syntax oneself, and participants work to ensure this coherency occurs. We have thus seen that speakers work for the syntactic suspension to emerge as a coherent and public pattern: when the display of strain or its after-effects is at issue, interlocutors contributions to the ongoing syntax would disrupt the socially relevant display.

The above analysis demonstrated how suspended syntax is accomplished and how participants manage potential incursions on the practice. In addition, syntactic suspensions can be used as a depiction (Löfgren and Hofstetter, 2021), in this case a depiction of strain where participants are not actually engaging in effortful activity but demonstrating what such activity looks like. Suspended syntax appears to be a conventionalized enough practice to be used as such a depiction of effort.

For example, in Extract 12 below, the director at the opera rehearsals depicts strain by suspending syntax in the middle of a quoted line from the libretto. The suspension is done as part of a depiction of the baritone’s character slowly moving into a kneeling position (line 7, see Figure 6).

At line 7, the director begins depicting someone bending down with difficulty, singing si la mia figlia “yes my daughter” in Italian, while embodying the character of the baritone. Between si “yes” and la (article), as well as between la and mia figlia “my daughter”, there are suspensions of the line’s progression (line 7), while at each suspension she moves her body a bit further towards the floor (see Figure 6). In addition, the director’s talk is produced throughout with a breathy voice quality. Together, these features serve as a depiction of how, for the character of the baritone, who is physically hindered, the descent towards the ground is causing effort (see Syntactic disruptions as a sign of bodily concerns on bodily concerns and syntactic suspension). The director herself is able to get down to the floor without difficulty, as seen in other moments in the data. The strain is thus imaginary but made available as something the baritone can reproduce when performing in character.

Temporal syntactic suspensions can thus be used to depict strain even in the absence of local physical effort. The suspension accomplishes the display, representing the out of breath or strained state of the body. Syntactic suspensions, as the multimodal practice described in this study, are thus usable to depict strain, even conventionalized enough for performances on a stage. This is not to say that all suspensions are conventionalized to depict strain, but that the practice of suspending syntax has one use that is highly recognizable as strain-related.