Pivots revisited: Cesuring in action


 The term “pivot” usually refers to two overlapping syntactic units such that the completion of the first unit simultaneously launches the second. In addition, pivots are generally said to be characterized by the smooth prosodic integration of their syntactic parts. This prosodic integration is typically achieved by prosodic-phonetic matching of the pivot components. As research on such turns in a range of languages has illustrated, speakers routinely deploy pivots so as to be able to continue past a point of possible turn completion, in the service of implementing some additional or revised action. This article seeks to build on, and complement, earlier research by exploring two issues in more detail as follows: (1) what exactly do pivotal turn extensions accomplish on the action dimension, and (2) what role does prosodic-phonetic packaging play in this? We will show that pivot constructions not only exhibit various degrees of prosodic-phonetic (non-)integration, i.e., differently strong cesuras, but that they can be ordered on a continuum, and that this cline maps onto the relationship of the actions accomplished by the components of the pivot construction. While tighter prosodic-phonetic integration, i.e., weak(er) cesuring, co-occurs with post-pivot actions whose relationship to that of the pre-pivot tends to be rather retrospective in character, looser prosodic-phonetic integration, i.e., strong(er) cesuring, is associated with a more prospective orientation of the post-pivot’s action. These observations also raise more general questions with regard to the analysis of action.


Introduction and background
Examination of language use in naturally occurring spoken interaction inspires a range of inquiries. One of them concerns the interface between syntax and prosody, particularly the alignment of syntactic and prosodic boundaries (see, e.g., Ford et al. 2002;Selting 1996;Selting 2001). While in many turns-at-talk, prosodic units closely correspond to syntactic units in that their boundaries are coterminous, in other cases speakers produce utterances where the alignment of these boundaries is, in the terms of this special issue, "fuzzy" (see, for instance, . One example of this occurs when speakers produce what Schegloff (1979, 275-6) originally termed a "pivot".
In syntactic terms, a "pivot" is a component of a turn used to both complete a first syntactic unit while simultaneously launching a second. Consider line 8 of the following extract (1), previously examined as a pivot by Walker (2007Walker ( , 2219 and Clayman and Raymond (2015, 389). Note that in this example, and in the others presented in this analysis, the pivot element is bolded (for other transcription conventions, see Appendix).¹ (1) "Bone" (NB 023/IV.3, sec. 171-174) ((AE telephone conversation between two sisters called Emma and Lottie. They are talking about dresses and shoes that they saw while shopping. The bone refers to the color of a pair of shoes.)) Syntactically, line 8 is parseable in two ways: "but I'd love the bone" and "the bone was so beautifu(l)". That is, "the bone"the pivot elementforms a first potential syntactic unit (A) with the material that precedes it (pre-pivot) and a second syntactic unit (B) with the material that follows it (post-pivot) (see Walker 2007;Norén 2013), as shown in Figure 1.
In prosodic-phonetic terms, all parts of the pivot pattern are generally said to be "through-produced", i.e., strongly integrated (Clayman and Raymond 2015, 388).² This prosodic-phonetic integration is typically achieved by prosodic-phonetic matching of the construction parts so that prosodic-phonetic features associated with major prosodic boundaries, and with potential turn-finality, are absent at the boundaries between pre-pivot, pivot element, and post-pivot. At these junctures there are neither low falls or high rises, nor final lengthening, loudness diminuendo, pausing, aspiration, (supra)glottal occlusions, etc. (Walker 2007). This is also illustrated in example (1), in which the pivot construction as a whole is through-produced prosodically-phonetically. There are no audible prosodic-phonetic boundary features before or after the pivot elementalthough, as we will show below (see Section 3.1), the pivot construction cannot quite be classified as a prototypical intonation unit (IU) either (see, e.g., Cruttenden 1997, 35). What is most relevant for our argument here is that not all instances of what have been treated under the heading of pivots and pivot constructions in previous research are actually that smoothly "through-produced". Closer examination of individual cases in fact reveals varying degrees of prosodic-phonetic integration in the production of such turnsranging from smooth through-production to rather separate IUs, as shown in example (2): (2) "Warm" (NB 006/I.6.R, sec. 55-58, from Clayman and Raymond 2015, 392) ((AE telephone conversation between two sisters called Emma and Lottie. They are talking about the current weather conditions.)) Here, the pre-pivot alone already forms a complete IU, as it has a focus accent (WA:RM) and a pitch contour which ends with a clearly final fall, followed by a pitch up-step into the candidate (B) part, among other features (for a more detailed discussion of this case, see Section 3.2). In addition to the stronger prosodic-phonetic separation of the potential pre-pivot and the pivot element, it should also be noted that syntax itself appears to enforce a right-bound interpretation of the don'tchu think as soon as the post-pivot is produced,³ thereby raising questions as to how such cases should be classified vis-à-vis instances such as (1). For this reason, Clayman and Raymond (2015, 392) use the term "pivot construction candidates", which we adopt here, in an effort to capture the contingent potential of such turns to be heard as "pivots", even if only momentarily.
In addition to the syntactic and prosodic packaging of pivots, it is also important to consider what such constructions are used to accomplish in social interaction. While it may initially be tempting to view these utterances as nothing more than "production errors", or the otherwise "ungrammatical" result of speaking off-the-cuff, research in Conversation Analysis and Interactional Linguistics (CA/IL; for overviews, see Clift 2016; Couper-Kuhlen and Selting 2018) strongly challenges such assumptions,⁴ showing pivoting to in fact be a practice that speakers mobilize for particular social-interactional purposes. One of the primary areas of research on pivot constructions concerns their use in the management of turn-taking (e.g., Walker 2007;Betz 2008;Clayman 2012;Clayman and Raymond 2015). Here they have been shown to be employed in the service of canceling the relevance of turn-transfer after the pivot element, thereby allowing the speaker to continue past a possible transition-relevance place (TRP), i.e., a point in talk where speaker change would be a relevant option (Sacks et al. 1974; see also Schegloff 2005a), as indicated by prosodic-phonetic, syntactic, and pragmatic completion points (e.g., . As a practice, pivots therefore cluster with a range of other turn-constructional devices that are deployed to this effect, such as rushthroughs (e.g., Schegloff 1982;Walker 2010), abrupt-joins (Local andWalker 2004), andincrementing (e.g., Auer 2006;Auer 2007;Couper-Kuhlen and Ono 2007;Ford et al. 2002;Schegloff 2016). Relatedly, pivots can be used in the context of repair, overlap management, and topic management (e.g., Betz 2008;Schegloff 1979; for a survey, see Horlacher and Pekarek Doehler 2014, 594).
With regard to the action(s) effectuated by pivot constructions, Clayman and Raymond (2015) state that the close prosodic-phonetic relationship of the construction parts "affiliates the pivot with both prior and subsequent talk and the action(s) they are implementing" (:401). The authors observe that "pivotal turn extensions do things in relation to [the] actions being implemented, [such as] […] modulating the action in progress" (:404, our emphasis, see also Horlacher and Pekarek Doehler 2014, 603) and that this may include "some form of elaboration of the action being implemented in the prior unit of talk" (Clayman 2012(Clayman , 1863emphasis in original). This work suggests that the close conjunction of the pivot parts in terms of their linguistic composition (i.e., prosodically-phonetically and syntactically) mirrors a close or "tight" relationship between those parts on the action dimension. At the same time, though, pivots have also been claimed to be usable in "refocusing or shifting to a different course of action" (ibid., our emphasis; see also  "hybrid turns" in Walker 2007Walker , 2235"perspective shifts" in Norén 2013). This diversity of observations in terms of what such turns are used to accomplish in talk calls for further work on aspects of the action(s) delivered by pivot construction candidates, in addition to, and in conjunction with, close examination of the prosodic-phonetic and syntactic delivery of such utterances.
The present study seeks to build on, and complement, prior research by exploring two issues in more detail: (1) what specific interactional work do pivot construction candidates accomplish on the action dimension (i.e., with regard to the action(s) involved in these turns' constitutive parts), and (2) what resources do participants bring to bear on the formation and recognition of these action relationships? In doing so, we take up Auer's (2006, 284) stance that the role that prosody plays in turn extensions has in many ways been undervalued (also Horlacher and Pekarek Doehler 2014, 596), and that turn extensions of various sortsincluding "pivots"need to be described more holistically, taking into consideration the syntactic, prosodic-phonetic, and action dimensions of the turn-at-talk in question (Auer 2006, 288).
In examining the prosodic-phonetic dimension of pivot construction candidates, we use the concept of cesuring (Barth-Weingarten 2016, also Barth-Weingarten and Ogden this special issue). Shifting the focus from "units" to "what comes between, and creates, them", and studying the latter with granularity, the term "cesura" refers to all sorts of discontinuities in the "flow of talk" created by prosodic-phonetic parameter changes, such as pitch up/down-steps, loudness and tempo changes, changes in voice quality, vocal-tract configurations, etc. Second, because these parameters are, by their very nature, gradient as opposed to binary, they can and do change to various extents: a cesura approach allows us to observe larger and smaller pitch movements, as well as greater and lesser degrees of loudness diminuendo, lengthening, and so on. Moreover, as the parameters are independent of each other, their changes can also cluster to a greater or lesser degree. This understanding is helpful because it solves a notorious problem of the IU model, namely how to deal with "fuzzy" IU boundaries (Cruttenden 1997, 35-7, for instance), in that it can easily capture prosodic-phonetic boundaries that are variously strong: IU boundaries, for instance, are usually considered to be characterized by a whole range of boundary criteria including final lengthening, pausing, aspiration, possibly a change in voice quality, as well as a change in pitch level and/or pitch direction and tempo (see, for instance, Cruttenden 1997, 35;see also Crystal 1969;Halliday 1985;Brazil et al. 1980;Brown et al. 1980;Du Bois et al. 1992, 100;Schuetze-Coburn 1994;Du Bois 2008), leading to categorization problems when some of them are missing (Auer 2010). Loudness diminuendo to zero dB without any other parameter changes, for instance, contextualizes only a held pause within an IU (see Local and Kelly 1986), while mere changes in vocal-tract configurations simply result in (the perception of) different phones (e.g., Ogden 2009). The "prosodic hierarchy" of the autosegmental-metrical approach already drew attention to variously strong unit boundaries (see, for instance, Nespor and Vogel 1986;Selkirk 1986;Beckman and Pierrehumbert 1986;Redi and Shattuck-Hufnagel 2001;Gussenhoven and Jacobs 2005). More recently, more discourse-oriented scholars have likewise shifted their attention more explicitly to "what comes between the units" (see also Peters et al. 2005;Mertens 2006;Degand and Simon 2009;Auer 2010;Barnwell 2013; see Barth-Weingarten 2016: Chapters 2.1 and 2.2 for a detailed survey of relevant literature on these various approaches). Yet, all these approaches still retain a dichotomy "IU"-"no IU" and "boundary"-"no boundary", respectively. The cesura approach, in contrast, radicalizes the focus on "what comes between the units". Unlike earlier approaches, it considers "units" to be mere epiphenomena and the (perceptual) result of more or less extensive (clusters) of prosodic-phonetic parameter changes, and it focuses on them rather than their results. This radical shift in perspective then also allows us to consider not only different kinds of "boundaries" of smaller or larger units, but all sorts of "breaks" in the flow of talk, regardless of whether they are created by smaller or larger (clusters of) changes in prosodic-phonetic parameters. In this way then, we can also systematically include weak(er) boundariesor rather "cesuras" to use the terminology of this special issue⁵ (see also Barth-Weingarten and Ogden this volume)into our investigation.
Incorporating this concept into our examination of pivots will allow us to show:


(1) how varied pivot construction candidates in English talk-in-interaction are with respect to prosodicphonetic packaging, i.e., their cesuring strength; (2) that this variability seems to map in an orderly fashion onto the action dimension in the relationship between the components of the pivot construction, with tighter prosodic-phonetic integration, i.e., weak(er) cesuras, being associated with post-pivot actions that are more retrospective or "backward-looking" vis-à-vis the action of the pre-pivot, and weaker prosodic-phonetic integration, i.e., strong(er) cesuras, being associated with more prospective or "forward-looking" postpivot actions; and (3) that both the prosodic-phonetic variation and the variation on the action dimension can be understood to form clines that parallel each other.
These observations, in turn, will lead us to some more general questions with regard to the formation and ascription of action in the immediacy of moment-by-moment talk (see Levinson 2013). The article as a whole, therefore, is also intended to serve as anadmittedly provocativeinvitation to bring novel perspectives to bear on the examination of turn design as it is used in the service of social action.

Data and methods
As a starting point for our data collection, we began with the examples already considered to be "pivots" in the relevant research literature on British and American English talk-in-interaction (Walker 2007;Clayman and Raymond 2015). For comparability and in accordance with CA methodology (see, e.g., Clift and Raymond 2018), we restricted ourselves to cases that cluster around one action/activity type which could roughly be referred to as "assessments/assessing". While not all of our examples are positively or negatively valenced "through the use of specific lexically assessing terms," as Thompson et al. (2015, 139, see also Li this special issue) define them, what our cases have in common is that a speaker is recognizably taking an evaluative stance toward some person, event, or inanimate entity in the target utterance (see, e.g., Goodwin 1986;Goodwin and Goodwin 1992).
We then expanded our collection with additional assessment pivot construction candidates from "classic" CA telephone corpora (e.g., Newport Beach, SBL) and individual recordings (e.g., TG, HGII; see Hoey and Raymond, in press) as well as some more recent corpora (i.e., CallHome, CallFriend; MacWhinney 2007). Candidate instances were identified on the basis of the syntactic criteria found in the literature: stretches of talk containing lexico-syntactic items that "can be understood as both completing the prior turn-constructional unit (TCU) and launching the next and hence a constituent of both units" (Clayman and Raymond 2015, 390; see also Walker 2007Walker , 2218, while we considered the exact nature of the candidate unit(s) involved (words, phrases, clauses) irrelevant (see also Walker 2007). In building our collection, we likewise did not consider as criterial the syntactic status of the pivot elementnamely whether it was syntactically obligatory or optional. Such an approach does not ignore the relevance of syntax, but instead allows us to examine empirically the ways in which syntactic structure may interact with prosodic-phonetic structure, and with action, as opposed to assuming a priori that syntactic categorization has a systematic effect across instances. This process generated a final collection of 26 cases, which formed the basis for this exploratory study.
To examine the data, we used the methodology of CA (e.g., Clift 2016) and IL (e.g., Couper-Kuhlen and Selting 2018) with their appreciation of the single case (Schegloff 1987, see also Barth-Weingarten and Ogden this special issue). We noted early on that there was a potential for circularity in our examination of these cases and our development of an argument. To avoid this pitfall as much as possible, we elected to conduct two rounds of semiindependent analysis, so as to keep our consideration of the prosodic-phonetic packaging of the pivot construction candidates (specifically, their cesuring features) separate from our analysis of the potential actions its parts could be seen to be implement, and vice versa. In what follows, we describe this research process.
First, two of the authors independently analyzed each case for the prosodic-phonetic design of the pivot candidate. Since we were interested in degrees of cesuring, this was first judged impressionistically and holistically (e.g., Local 2007, 3). This first holistic auditory impression was then complemented by a more detailed auditory analysis of the pivot construction candidates and their component parts. Apart from a coherent pitch contour and nucleus syllable (see Cruttenden 1997, 35;Du Bois et al. 1992, 100 for internal criteria for IUs), we focused in particular on audible cesuring parameter changes (see Section 1). These included final pitch movements following a focus accent, changes in voice quality (mostly creakiness), final lengthening, loudness decreases, the release of final plosives at the end of the first candidate unit, on the one hand, and unfilled pausing, tempo changes (latching, anacrusis), and pitch up/down-steps at the beginning of the second candidate unit, on the other, as well as the extent of all of these parameter changes and their degree of clustering (for details of this analysis, see Barth-Weingarten 2016, 168). In cases of doubt, the auditory analyses were checked with the help of the speech analysis software PRAAT (https:// www.fon.hum.uva.nl/praat/), which allows for the visualization of most of the relevant parameters. This resulted in some informal quantification of the parameter changes, i.e., "greater/smaller than" statements. Measurements were not taken. On that basis, the two analysts then ordered the examples on a cline of prosodic-phonetic cesuring, ranging from cases with only weak cesuras to those that were delivered with strong cesuras.
Concurrently, two of the authors (one overlapping with the prior pair) analyzed the action(s) accomplished by the (A) and the (B) parts of each pivot construction candidate. While action analysis must necessarily take into account the prosodic-phonetic design features of TCUs and turns-at-talk (see, e.g., Barth-Weingarten et al. 2010;Couper-Kuhlen and Ford 2004;Couper-Kuhlen and Selting 1996), it does not focus exclusively, or even specifically, on prosodic-phonetic parameter changes. We therefore opted for a "deconstructive-reconstructive"-style approach in our examination of the collection in terms of action: In a first step, we sought to identify possible actions (or action potentials) the target turn could be heard to implement in context. Its prosodicphonetic packaging was then considered in a second step, to determine whether and how it could be seen to bear on those potential action interpretations. Thus, far from ignoring the relevance of phonetics/prosody to the analysis of action, our research process was designed specifically to expose the import of such features of turn design. This pair of authors then discussed how the actions implemented by the parts of the pivot construction candidate related to each other and whether there were any systematicities in those relationships.
Finally, we collectively explored how the cases could be grouped/ordered with regard to their prosodicphonetic cesuring and action relationships, respectively. Our analysis suggests that the prosodic-phonetic and the action dimensions of talk are related, and in ways that are more detailed than has been appreciated in previous research.

Pivots revisited: Cesuring and action
In what follows, we build our argument by first considering the extremes of the prosodic-phonetic cline: the most and the least prosodically-phonetically integrated cases. We then move to consider a case that is situated in between these prosodic-phonetic extremes.
In order to give the reader a feel for how we went about the analysis, for each case, we begin with an examination of the syntactic and prosodic-phonetic features of the example, and then discuss the action(s) implemented by the turn. In the discussion, we take stock of the action relationships seen across this array of examples, and propose a cline on the action dimension that we argue maps onto the strength of prosodicphonetic cesuring. Specifically, we will show that while tighter prosodic-phonetic integration, i.e., weaker cesuring, co-occurs with a post-pivot action whose relationship to that of the pre-pivot tends to be rather retrospective in character, looser prosodic-phonetic integration, i.e., stronger cesuring, is associated with a more prospective orientation of the post-pivot's action.

Weak(er) cesuras: Redesigning an action-in-progress
One extreme of the prosodic-phonetic cline is constituted by cases that are indeed smoothly throughproduced, meaning that they exhibit only very weak cesura(s) across the construction, as we saw in example (1). In such cases, we find that, notwithstanding the absence of the classic harbingers of repair (e.g., cut-offs, hitches), speakers are nonetheless demonstrably working to redesign the in-progress action. That is, the speaker produces an action (A) and uses the pivotal element therein to shift into an alternative design of that action (B), thereby generating a turn that better fits the sequence of actions in progress (see Drew 2013;Drew et al. 2013;cf. Schegloff 1979). While this process of redesigning an in-progress action undoubtedly shares a family resemblance with self-repair operations like "replacement" (see Schegloff 2013), it is not clear to us that the (B) part of such cases is designed so as to be heard as "replacing" the (A) part of the utterance. Rather, the (B) part involves the speaker shifting to a design for the action-inprogress that is addressed to other, possibly competing exigencies of the sequence underway without allowing an interjacent TRP to occurhence our use of the term "redesign".⁶ Consider the following case (3).
(3) "Sarah McLachlan" (CallFriend 6093, sec. 200-220) ((AE telephone conversation. Bob is asking Sue for advice about what to get his crush for her upcoming birthday. Sue's proposala card and a CD that she likesposes the problem that Bob does not know what music his crush likes. Sue then starts listing possibilities for finding out and moves on to suggesting specific artists.⁷)) Sue's utterance in line 4 is syntactically parseable as "mine happens to be Sarah McLachlan" (A) and "Sarah McLachlan's really good" (B), with the proper name "Sarah McLachlan" serving as the pivot element. Syntactically, it is obligatory in both parts of the pivot construction candidate, since it acts as subject complement in the (A) part, while at the same time constituting the subject NP of the (B) part.
In prosodic-phonetic terms, the pivot candidate is smoothly through-produced. The pivot element carries the only focus accent of the two candidate units (-LACH-), which is thus shared by both construction parts. Other than the vocal-tract configuration changes necessitated by the lexical tokens, there are no parameter changes audible around it (for illustration and further evidence, see also Figures 2 and 3).  Raymond (2019, 195) synthesizes, "as distinct from "classic" cases of repair due to their attempt at preserving some element(s) of the putative trouble source, as opposed to endeavoring to replace it completely with the repair solution" (see also Jefferson 1974). 7 Note that the participants in this call are bilingual Spanish-English speakers, hence Sue's momentary code-switch to Spanish in line 8. Her ay chico no can be translated as "oh boy no".
With regard to action, the pivot construction candidate here contributes to an on-line redesigning of the action-in-progress. Prior to line 4, Sue has been giving Bob advice as to how to figure out what music his crush likes. With the pivot utterance in line 4, she begins to take a different tack by naming artists whose music she (Sue) likes very much and which may therefore stand as a more specific recommendation of what CDs Bob could get for his crush. While the initial design of this recommendation (mIne happens to be sarah macLACHlan, line 4) ties back to Sue's prior advice to talk to his crush about what her best favorite ceeDEE is (line 3), Sue subsequently treats such a uniquely subjective framing as sub-optimal for recommending a gift for another individual. This is reflected in her use of the pivot to rework precisely this aspect of the design of her utterance-in-progress. With it, she shifts from what Edwards and Potter (2017) call an S-side (a "subjectively" framed) assessment to an O-side (an "objectively" framed) assessment. The O-framed design  of this assessment demonstrably contributes to the action of recommending artists that anyone could like, since it less strongly foregrounds Sue's personal preferences, but asserts an "objectivized" stance (see also Shaw et al. 2015). Note that Sue retains this O-frame when subsequently embarking on further recommendation-relevant assessments (e.g., (Björk) is not BAD, line 6). In this example, then, we see a speaker using a pivot construction to redesign an in-progress action, thereby gesturing toward self-repair but without the progressivity-halting features typically associated with a self-repair operation (Schegloff 2013). The prosodic design, apart from not providing any space for incomings, supports this also by "passing over" the candidate repair with a format that does not exhibit any cesura and is thus unmarked (see Plug 2015 for prosodically (un)marked self-repair).
For another example of this sort, let us return to example (1), reproduced below as (4) with more sequential context. In prosodic-phonetic terms, it exhibits slightly more cesuring features than example (3), just discussed, but it is still tightly integrated.
(4) "Bone" (NB 023/IV.3, sec. 164-180) ((AE telephone conversation. Emma and Lottie are sisters. They are talking about dresses and shoes that they saw while shopping. The bone refers to the color of a pair of shoes. Earlier in the conversation, Emma had already recommended that Lottie acquire them, suggesting they would be a good match for one of her dresses, but Lottie has deflected this recommendation.)) As described briefly in the introduction, line 8 is syntactically parseable as "but I'd love the bone" (A) and "the bone was so beautif(ul)" (B). Again, the candidate construction parts share a common constituent, the pivot element "the bone", which serves as the direct object of the (A) part and the subject of the (B) part.
In prosodic-phonetic terms, the pivot candidate is less smoothly through-produced than example (3) but still very tightly integrated. This is due to it containing not just one, but two rising-falling pitch contours (Figure 4). Moreover, the accent patterns of the (A) and (B) parts mirror each other (a secondary accent followed by a more prominent one). Of the prominent syllables, BEAU-turns out to carry the focus accent in the end. But from an on-line perspective, bOne at the potential completion point of the (A) part could also qualify as the main accent. However, the intensity decrease stops there, and the syllable lengthening and falling pitch contour continue on into the post-pivot. Moreover, bOne is also lower in pitch than the first high-pitched accent of the (A) part (lOve). In effect, then, it becomes just another secondary accent in the pivot construction candidate. In addition, in the two mirroring sub-contours, the extra-high intensifier ↑sO stands out so much that it can be heard as preparing the ground for the actual focus accent BEAU-another way of tying the candidate construction parts together.⁸ Moreover, and  8 In addition, the so projects lexico-syntactically. more importantly for our current purposes, no other cesuring features other than the vocal-tract changes necessitated by the lexical tokens are audible before or after the pivot element (see also Figure 5). If we stop the audioand thus artificially divide the candidate construction before and after the pivot element, respectivelyboth partseach including the pivot elementsound like any normal IU with a coherent contour, a focus accent, final lengthening and loudness diminuendo, except that the final lengthening of the (B) part is slightly compressed for a rush-through into the next TCU. In sum, then, the pivot construction candidate here is not quite as smoothly through-produced as in example (3), and is thus positioned not quite at the extreme of our prosodic-phonetic cline; but it still only exhibits rather weak cesuras.
With regard to action, we see an action relationship between the (A) and (B) components of the construction that is very similar to the one observed in example (3) above. After the prior sequence has  come to an end, Emma brings up "the bone" again with what ends up being the pivot construction candidate. We know from prior talk (data not shown) that Emma had already made an earlier attempt at convincing Lottie of the virtues of the bone shoes, so the but-preface may not only mark a contrast with the negative assessment in line 3, it may also serve as a "resumption" marker (Mazeland and Huiskes 2001). The pivot construction candidate as a whole then delivers an assessment of the color of the shoes, but again the two parts offer different types of assessments: an S-side assessment in the (A) part, and an O-side assessment in the (B) part (Edwards and Potter 2017). So again, we see an on-line redesigning of the actionin-progress and, notably, another shift from an S-side assessment to an O-side assessment.⁹ As in example (3), it appears that this shift is done in the service of fitting the assessment to Emma's project of recommending that Lottie acquire the shoes: an assertion of the "objective" beauty of the shoes is constructed as an integral part of the recommendation action, and indeed may be more persuasive than the fact that Emma personally loves them (the dense accentuation on the intensifier and the assessing term sO: BEAUtifu(l) can be seen to place further emphasis on this part of Emma's assessing turn).¹⁰ The O-side assessment presents "acquiring the bone-colored shoes" as being in Lottie's interest; an analysis that is further supported by Emma's subsequent "other-oriented" assessments in lines 12 and 14, which can be heard to further this agenda. Note that Lottie unambiguously hears Emma's turn this way by responding with reluctance: Her YEA:H in line 11 is delayed and produced with slightly rising pitch, and her YEAH in line 16 is overlapped and truncatedboth turns thereby being audibly uncertain and noncommittal (Thompson et al. 2015). In sum, then, the pivot in this case is again used to redesign an action-in-progress on the fly, so as to adjust aspects of the turn's construction to the purpose at hand, without allowing an interjacent transition-space to occur.
With the analysis of these cases in mind, let us now move to the other end of our prosodic-phonetic cline.

Strong(er) cesuras: Occasioning a response for an action
In contrast to the cases discussed in the previous section, in which there were only weaker cesuras in the target utterance, the cases we present in this section exhibit a range of prominent prosodic-phonetic parameter changes at specific syntactic junctures of the pivot construction candidatei.e., they contain strong(er) cesuras. At the same time, in terms of action, we find that the (B) part of pivot construction candidates with more cesuring features does not work to redesign an in-progress action, but rather to implement an action that works to occasion a recipient response for the action of the (A) part (e.g., when the current speaker projects that recipient uptake for the turn-so-far could become an issue).
As a first exemplar of this sort, reconsider example (2), which we repeat below as example (5) with more sequential context. As described in the introduction, this case illustrates a pivot construction candidate with a rather strong cesura between the two parts.
 9 According to Potter et al. (2018Potter et al. ( , 2020, shifts from S-side to O-side assessments (like the ones observed here) are markedly less common than shifts in the opposite direction. Moreover, the authors note that in cases of such a shift, where it is done in separate TCUs, the first part (the S-side assessment) often already gets some sort of uptake at the first TRP. In our view, this constitutes converging evidence for the analysis forwarded for examples (3) and (4)i.e., that this shift relevantly informs the use of a pivot to redesign the in-progress action without allowing an interjacent transition-space to occur. 10 It is noteworthy, however, that unlike in example (3), where Bob has been soliciting advice from Sue, in example (4) Emma is volunteering a recommendation via an advice-implicative assessment (see Shaw et al. 2015). Given the limitations of our collection, we cannot pursue the issue of whether this difference can be seen to have a bearing on aspects of turn-design in more detail, but we mention it as a possibility to be explored in future research.
(5) "Warm" (006/NB.I.6.R, sec. 51-59) ((AE telephone conversation. Lottie and Emma are talking about the current weather conditions.)) In theory, the pivot construction candidate in lines 4-5 is syntactically parseable as "Oh it's warm, don't you think?" (A) and "Don't you think it is?" (B). However, in retrospect, the continuation with it I:S enforces a right-bound interpretation of the candidate pivot element, and the prosodic-phonetic phrasing with a strong cesura between WA:RM and don'tchu (lines 4-5) underscores this. As can be seen from Figure  6, the strong cesura is created by the falling-to-low intonation on the focus accent WA:RM (line 4). In addition, the focus accent exhibits extra length, there is a loudness diminuendo and a change in pitch direction into a next contour. The latter is then not only different in kind (simple rise vs. rise-fall in the first chunk) but also has a pitch range which equals that of the possible pre-pivot, so there is no prosodic subordination (see, for instance, Couper-Kuhlen 1986, 106). The only "integrating" features that are left here are due to the rushing into the next unit ("pushing", Schegloff 2005a): (1) the final bilabial nasal [m] in WA:RM is slightly compressed, see also the latching, i.e., fast movement into the next chunk, and (2) the voiced alveolar plosive [d] does not exhibit full closure of the vocal tract (Figure 7).
The two parts of the pivot construction candidate also implement relatively distinct (albeit closely related) actions. The (A) part offers an assessment of the weather, whereas the (B) part invites agreement with it. With its rising intonation and interrogative morpho-syntax, the (B) part in this case exhibits strongly response-mobilizing features (Stivers and Rossano 2010). Moreover, we can observe that Lottie adjusts the terms of Emma's first assessment from second position. Note that Emma's first assessment is negatively formatted (it's not too cold, line 3), which, as Schegloff (2005b, 455) observes, "is a way of invoking the possibility of complainability". Lottie, however, responds with a positively formatted second assessment (oh it's warm, line 4). Her response is therefore hearable as more than a simple upgraded agreement in second position (Pomerantz 1984). In fact, it is implicative of disagreement or other-correction (note also the oh-preface, Heritage 2002). The following (B) part of the pivot construction candidate (don't you think it is?, line 5) then strongly (re-)invites agreement with this "revised" counter-assessment and makes it come off more like a new first assessment than the second assessment that it is by virtue of its sequential position (on which, see Heritage and Raymond 2005). This "distinctness" of the two parts of the pivot construction candidate on the action dimension is reflected in the strong prosodic-phonetic cesura between them.
A second case that illustrates this sort of action relationship can be seen in example (6). In prosodicphonetic terms, the two parts of the pivot construction candidate are not separated by as strong a cesura as that in the prior example (5), but the break in the flow of talk is still well audible.
(6) See that house (NB 028/IV.10.R, sec. 91-103) ((AE telephone conversation between two sisters, pseudonymed Emma and Lottie. Lottie has just returned from a visit to a mutual acquaintance who lives in Palm Springs. The following sequence resumes an earlier, appreciative recount of that trip.)) Syntactically, the address term (e_h_mma, line 4) is the pivot element, yielding "you should see that house Emma" and "Emma you have no idea" as the (A) and (B) parts of the pivot construction candidate, respectively. Clayman and Raymond (2015) discuss address terms as an example of a resource that can serve as a "modular pivot"i.e., a generic, syntactically optional pivot candidate, which "can be used to extend a wide range of sentential units" (:390) and "may or may not function as actual pivots on any particular occasion," depending on how they are realized phonetically (:392).¹¹ In prosodic-phonetic terms, the (A) part could stand as a complete IU with the focus accent on HOUSE and an overall falling contour. Apart from the necessary vocal-tract changes, we find no pitch, tempo, loudness etc. changes, nor does e_h_mma start with a glottal stop (see also Szczepek Reed 2014). Instead, it continues the falling contour ( Figure 8) and is therefore also too low in pitch to count as the onset of the next chunk, especially as that is another falling contour. In effect then, there is only a very weak cesura before the pivot element, nothing like an IU boundary. After the pivot element, in contrast, the pitch moves audibly up, even if that is only fully the case on the vowel of the next lexical item, the pronoun "you". Nevertheless, the end of the (A) part and the post-pivot -e_h_mma and youare also integrated by assimilationthe palatal approximant/semi-vowel of "you" ([j]) already starts during the final parts of the offset vowel of "e_h_mma" (Figures 9 and 10, note in particular the unusually low first and high second formants, indicating the frontedness of the "a")and an increase in tempo that starts on the pivot element, i.e., before the potential syntactic completion point that follows e_h_mma. Moreover, the candidate units are audibly integrated in a declination unit (see Schuetze-Coburn et al. 1991;also "supradeclination" in Cruttenden 1997). Hence, the prosodic-phonetic cesura is not as strong as in example (5), but it is still clearly audible.
On the level of action, as in the prior case (5), the (B) part is used to occasion a response for the action delivered in the (A) part. Lottie's deontically framed assertion (you should see that house (Emma), line 4) has the pragmatic force of a positively valenced, exclamatory assessment. Her subsequent (Emma) you have no idea serves to foreground the participants' differential epistemic access and can therefore be heard to upgrade  11 Note that the syntactic optionality of address terms (and other modular pivots) allows for such turns to be parsed in a third way, namely with the address term acting as an independent phrase: "you should see that house" "Emma" "you have no idea." However, to be hearable as such, this item would have to be realized as a clearly separate IU (i.e., with no integrating features whatsoever to either side), and such cases were excluded from the present analysis. This is due to the fact that an address term produced in this way loses its ability to serve as a potential pivot between two possible TCUs around what would otherwise be recognizable as a TRP; that is, it no longer acts as a pivotal turn extension. Rather, such a turn would consist of three separate TCUs, with two intervening TRPs between them. the assessment and its newsworthiness (Heritage 2012a, 4). More specifically, it offers a direct description of Emma's state-of-knowledge, an incursion into the recipient's epistemic territory that prior research has shown to be response-mobilizing (Stivers and Rossano 2010;Heritage 2012b). Indeed, Emma orients to these two facets of Lottie's turn, which can be seen in the details of how she constructs her response. She first aligns with the epistemic asymmetry indexed in the (B) part of Lottie's pivot utterance, issuing agreement with i bet it's a DREAM (line 5). She then orients to the assessment's newsworthiness by asking a topicalizing question in line 6, which is grammatically built to attach to the (A) part of the pivot construction candidate (i.e., You should see that house […] with the swimming pool enclosed, huh?)¹² and leads to further talk about Lottie's trip. Here, then, as in example (5), the action in the (B) part of the pivot construction candidate works to occasion a response for the action in the (A) part by instantiating a steeper epistemic gradient, which in this case works sequentially to occasion further discussion of Lottie's trip (see also Heritage 2012b).
At the strong(er)-cesura end of the prosodic-phonetic cline, then, we see that the two parts of the pivot construction candidate do recognizably distinct work on the action dimension, with the (B) part being used to prospectively follow up on the action of the (A) part, occasioning a response to it. This action relationship is distinct from that seen in the prosodically-phonetically integrated cases discussed in the weak(er)cesura section, where the (B) part was deployed to redesign the speaker's emergent action.
Thus far, we have presented cases that are at, or close to, the two extremes of our prosodic-phonetic cline: cases with very weak cesuras as well as cases with strong(er) ones. There are also cases in our collection, however, that fall in between these two extremes.

Intermediate cases: Elaborating an action with a subsidiary action
Case (7) below is an example of a case that exhibits stronger cesuras around the pivot element than the integrated cases (Section 3.1); at the same time, the parameter changes are not as many or as extensive as Figure 9: Zoomed-in PRAAT sound pressure waveform with spectrogram of the pivot candidate in example (6), line 4 (the equal ("=") sign marks the assimilatory juncture between e_h_mma and you [see Figure 10 and footnote 13 for a more detailed description]).


12 The significance of this continuative syntactic design becomes clear if it is compared to alternative formatting options that would have contextualized Emma's question as an independent, next action (e.g., (And) it's got the swimming pool enclosed, huh? or Does it have the swimming pool enclosed?). This suggests that Emma's turn is carefully designed so as to engage with both parts of Lottie's pivot utterance, while orienting to the preference for contiguity (Sacks 1987). they were in the cases with strong(er) cesuras (Section 3.2). Moreover, with regard to action, we find that the (B) part of the pivot construction relates to the (A) part not by redesigning the action-in-progress, or by producing an action that seeks to occasion a response to the prior. Rather, in examples such as (7), the (B) part of the pivot construction appears to constitute what Auer (2007, 649-50) has referred to as a "retrospectively oriented, subsidiary" action. Such actions are subsidiary in that they refer back to and elaborate the previous action, with "specifications" constituting one routine example (ibid., Horlacher and Pekarek Doehler 2014, 609). Consider example (7). For the purposes of the present discussion, the pivot construction candidate in line 7 can be parsed in two ways (but recall footnote 11): "you don't look it, Jen" (A) and "Jen, I must be honest" (B), both of which share the term of addressagain a modular pivotas the pivot element.  13 Note that immediately on release of the nasal consonant (at about 0.13 s in Figure 10), F2 is somewhere around 1,850 Hz, which is quite high for the kind of vowel at the end of Emma's (real) name. Its trajectory is upwards, to around 2,270 Hz at the peak of [j] (at around 0.26 s in Figure 10). This indicates that Lottie's tongue is already fronting towards [j] at the point at which the nasal airflow is stopped and released into the final vowel of the address term. Moreover, the low F1 at about 360 Hz also suggests an early move into a close/high articulatory setting that is consistent with [j]. We would like to thank Richard Ogden for assistance with the spectrogram analysis.
In terms of prosodic-phonetic integration, the two construction parts in line 7 are neither tightly, nor weakly, integrated. In retrospect they could be perceived as two IUs as there are two candidate focus accents (LOOK, HONest)one in each of what are perceivable as two similarly steadily falling pitch contours (see also Figure 11). Moreover, these potential IUs are separated by pitch movements: a down-step toward the end of the (A) part and an up-step slightly after the beginning of the (B) part. However, locating the cesura between them is rather difficult, especially when we consider the participants' on-line perspective: the first potential focus accent LOOK is followed by a slight pitch down-step leading to the lower pitch of the candidate IU tail/TCU ending of the pre-pivot, which the modular pivot element candidate jen continues. jen also sounds a bit too long for a potentially anacrustic onset syllable, though not long enough for final lengthening (Barth-Weingarten 2016, 168). Only the following I is produced with clear, and thus potentially TCU-initial, anacrusis. Yet this lexical item still continues the previous lower pitch level. We can only hear a pitch up-step on mustcoming later than the syntactic completion point (Clayman and Raymond 2015, 392). In effect, then, we are dealing with a cesural area (Barth-Weingarten 2016, 75-6) that embraces "it jen I"i.e., the end of the pre-pivot, the modular pivot element, and the beginning of the postpivot. Cesuring features are spread across a stretch of talk, more so than in example (6). Moreover, the pitch up-step is also smaller than the up-step in example (6). In addition, the (A) and (B) parts are also integrated by the fact that their potential focus accents occur on one falling declination line with peaks at 303 and 252 Hz.¹⁴ In effect, then, this pivot construction candidate stands between the cases with weak(er) cesuras discussed in Section 3.1 and the ones with strong(er) cesuras discussed in Section 3.2. The cesura between them is weak(er) in kind and fuzzier in position.
On the action dimension, too, this case appears to fall in between what was observed in the examples analyzed in the previous two sections. Here the (A) part forwards a reaffirming positive assessment as a compliment, whereas the (B) part offers an avowal of truthfulness. This avowal serves to bolster up the credibility of the preceding compliment (Clayman 2012;Edwards and Fasulo 2006), arguably so as to avoid a possible "pro forma" hearing of it based on sequential grounds. As can be seen, Jen has self-interruptively launched a news announcement with self-deprecation implicative news in lines 1-3. This is followed by a bit of delay (line 4), which may be due to the sudden, abrupt, and sequentially disjunctive character of Jen's  14 Unfortunately, the sound quality of this particular example is too poor to reliably detect any glottal closures around the modular pivot and to produce a spectrogram. news announcement. What follows the delay is a first compliment (well now you don't LOOK it, line 5) that seeks to deflect the self-deprecation embedded in Jen's news announcement by rejecting its visible noticeability (see Pomerantz 1984). However, because of the preceding delay, this compliment is potentially hearable as "pro forma". Moreover, it overlaps with Jen's prosodically upgraded repetition of tEn STO:NE in line 6, which upgrades the news value of, and expresses her exasperation at, the weight gain. In this context, Ann's mere repetition of "you don't look it (Jen)" may not be sufficient to deal with Jen's overlapping upgrade. That is, it may come off as too weak and thus potentially "pro forma" again. Her subsequent avowal of truthfulness in the (B) part of the pivot construction works to handle this contingency.¹⁵ In this example, then, the (B) part of the pivot construction is clearly not used for an on-line redesigning of the action-in-progress as with the cases in Section 3.1. Indeed, the (B) part ("(Jen) I must be honest") in this case is pragmatically dependent on the prior (A) part ("You don't look it (Jen)") and would thus simply not work interactionally without it. Likewise, the (B) part does not issue a distinctly responsemobilizing action for the action issued by the (A) part as with the cases in Section 3.2. That is, although the recipient does provide a response (a mild disagreement), the (B) part of Ann's pivot turn is not designed so as to mobilize or occasion such a response, at least not in the same way as seen in the (B) parts of those earlier cases (a point to which we will return in the next section). Rather, the pivot construction in this case is better understood as implementing a subsidiary action that retrospectively elaborates the main action in the preceding (A) part (Auer 2007, 649-50). The elaboration here works to buttress a prior evaluation that is vulnerable to a "pro forma" hearing and thus potentially doubtfula usage pattern well-documented for pivotal turn-extensions done with address terms (see, in particular, Clayman 2012Clayman , 1863. The prosodicphonetic design nicely parallels this "in-between-ness" on the action dimension.

Summary and discussion
In this article, we considered instances from the extant literature on pivots, as well as additional pivot construction candidates, to examine in greater detail their prosodic-phonetic design and the action(s) they are used to accomplish. Adopting the cesura approach (Barth-Weingarten 2016; Barth-Weingarten and Ogden this special issue), we have shown that pivot construction candidates do not only include smoothly "through-produced" cases (Clayman and Raymond 2015, 388), but that they exhibit varying degrees of prosodic-phonetic integration. They can thus be ordered on a continuum that ranges from weak(er) cesuras (Section 3.1) to cases with strong(er) cesuras (Section 3.2), with intermediate casesin which the possible parts of the pivot candidate are neither tightly integrated nor bordering on separate IUs, i.e., designed with cesuras that are of intermediate strength and fuzzy in charactersituated between these extremes (Section 3.3).
We furthermore illustrated that these different cesuring strengths are associated with distinct action relationships between the (A) and (B) parts of the construction as one moves across the prosodic-phonetic cline. We argued that in the cases with weak(er) cesuras, speakers are engaged in redesigning their inprogress action (Section 3.1). At the strong(er)-cesura end of the cline, by contrast, speakers use the (B) part of the construction to issue an action that works to occasion a response for that in the (A) part (Section 3.2). In between, we find cases in which the (B) part implements a subsidiary action that retrospectively elaborates the main action delivered in the preceding (A) part (Section 3.3; Auer 2007). This patterning suggests that, at least as far as pivot construction candidates are concerned, the prosodic-phonetic and the action dimension are connected in more intricate ways than has thus far been appreciated in the existing literature.
In line with prior research on pivots, we have, up to this point, discussed these action relationships in a largely categorical fashion. However, the clearly scalar nature of the prosodic-phonetic delivery of pivot constructions invites consideration of the possibility that the action relationships just described may  15 That Jen pushes back yet again in her response may suggest an orientation on Jen's part toward having heard Ann's deflecting compliments as possibly "pro forma". likewise be arranged on some sort of continuum. One dimension of action that seems relevant in this regard is that of retrospectivity vs. prospectivity. We posit that the (B) part of pivot construction candidates relates to the (A) part on a cline from more retrospective (or "backward-looking") to more prospective (or "forwardlooking") in terms of the action(s) they implement.
That we are indeed dealing with a cline on the action dimensionand not discrete categoriesis supported by the fact that additional subdivisions can be made between our cases along this continuum of retrospectivity to prospectivity. Consider once more the two cases that we discussed categorically as the (B) part issuing a response-mobilizing action (Section 3.2): example (5) "Oh it's warm don'tchu think it is" and example (6) "You should see that house Emma you have no idea." While the (B) part in both of these examples is generally prospectively oriented in that it seeks to occasion further talk from the recipient, the two cases are at the same time slightly distinct in just how much pressure they exert in doing so (Stivers and Rossano 2010). With its incorporation of interrogative morphosyntax and turn-final rising intonation, the (B) part in example (5) "Oh it's warm don'tchu think it is" incorporates more response-mobilizing features (ibid.) and is thus arguably more prospective in its orientation than the declaratively-framed, falling-to-low (B) part in example (6) "You should see that house Emma you have no idea", where the response-mobilization is largely epistemically driven. Conversely, the (B) part in "You should see that house Emma you have no idea" is at the same time more retrospectively oriented than the (B) part in "Oh it's warm don'tchu think it is". The latter ("don'tchu think it is") is almost exclusively "forward-looking" and can be conceived of as a more elaborate version of a tag-question, deployed to exert response-relevance. By way of comparison, as was detailed above, the (B) part in "You should see that house Emma you have no idea" serves to adjust the epistemic gradient indexed in the (A) part. And in this regard, it also resembles intermediary cases like example (7) ("You don't look it Jen I must be honest") to a certain extent, in which the (B) part retrospectively elaborates the preceding (A) part with a subsidiary (e.g., supportive) action.¹⁶ Importantly, it is not merely that a cline on the dimension of action can be argued to exist, but that this cline parallels the cline we see on the prosodic-phonetic continuum. As detailed earlier, "Oh it's warm don'tchu think it is" exhibits more parameter changes, i.e., a stronger cesura, than "You should see that house Emma you have no idea" does. Notwithstanding the categorical generalizations used in the body of this article (on which, see Schegloff 1993), then, it is clear that continua exist with regard to both the prosodic-phonetic packaging and in terms of action, at least with regard to the particular dimension of retrospectivity vs. prospectivity of the post-pivot's action relative to the pre-pivot's action.
To summarize our findings, then, the prosodic-phonetic and action clines that we propose grossly map onto one another in that the maximally retrospective "redesigning an in-progress action" pivots are rather smoothly integrated, the retrospectively elaborating pivot construction candidates exhibit intermediary degrees of prosodic-phonetic integration, and pivot construction candidates with rather prospectively oriented (B) parts show the least integration. These two clines can be visually represented as in Figure 12.¹⁷ The scalar account we offer here is supported by prior work in myriad ways. First, our analysis finds common ground with prior explorations of turn-taking, transition relevance, and the formation of TCUs in the service of social action: If one TCU implements one action, we would expect backward-looking pivot constructions that "redesign" an action-in-progress, to be more tightly integrated prosodically-phonetically, thereby more closely resembling "one" TCU in terms of its prosodic-phonetic packaging. By contrast, we would expect pivot constructions that implement distinct actionse.g., one primary action followed by a follow-up action that mobilizes responseto exhibit a stronger cesura, thereby approaching separate TCUs in terms of their prosodic-phonetic packaging. This is precisely what we find in the examination of our collection. In a similar vein, our analysis also squares with earlier studies showing that  16 Note again, however, that the (B) part in "You don't look it Jen I must be honest" (example 7) cannot be said to be prospectively oriented in the same way that the (B) parts in examples (5) and (6) can (i.e., as being used to mobilize a recipient response). 17 While we do not mean to reinvoke the categorical analysis that we have set aside in favor of a scalar understanding, or ignore the sequential particulars of each case, it bears mention that the cases in our collection are distributed roughly evenly across the two clines. That is, we could not observe any remarkable distributional skewings toward either end of the continuum. comparable relationships hold between the prosodic-phonetic design of successive turns-at-talk and the actions they implement at the level of sequence, with more disjunctive prosodic-phonetic designs indexing sequential disjunctiveness as well (e.g., Goldberg 1978;Couper-Kuhlen 2001. We see the same basic mechanism in operation herenot between successive turns in a sequence, but rather within the emergent design of turns-at-talk that incorporate pivots. As we have demonstrated, the greater the prosodic-phonetic disjunction between the constituent parts of pivot constructions, the more they approach independent TCU status and, correspondingly, begin to have their own distinct import as actions, both pragmatically and sequentially.¹⁸ Finally, and on a more general note, it bears mention that this analysis also parallels other recent scalar reconceptualizations of longstanding interactional phenomena in CA/IL, such as (1) conditional relevance and the mobilization of response (Stivers and Rossano 2010;cf. Schegloff and Sacks 1973), (2) the organization of preference (Robinson 2020), (3) the entitlement-contingency cline of request formulations (Curl and Drew 2008;Raymond et al. 2021), and (4) continua related to recruiting the assistance of others (Kendrick and Drew 2016), among others.
Notwithstanding the contributions that we hope to have offered through this report, it bears emphasis that our analysis has been designedly exploratory in nature. As such, while we were purposeful in our decision to limit the examination to a relatively small number of cases, all with a similar action agenda (Section 2), future research is necessary to determine whether and to what extent our account can be applied to other action types that routinely incorporate pivot constructionsthe most frequent (after assessing) being "enquiring" and "reporting", according to Walker (2007Walker ( , 2217. Comparing and contrasting action environments will undoubtedly lead to a more refined operationalization of concepts like retrospectivity and prospectivity, as well as the gradient between them, as relevant features of how actions can relate to each other. Indeed, it may reveal other (scalar) dimensions of action relationships that are likewise relevant to the particulars of how pivot constructions are produced and understood in context. We can offer two hypotheses in this regard: On the one hand, because different action types implicate different interactional exigencies (e.g., making a delicate request vs. telling a funny story), there may be different orders of relevance for pivoting as a turn-constructional practice as we look across categories of action. On the other hand, though, as a turn-constructional practice, pivoting essentially serves to bind otherwise separate TCUs together in some fashion, and so indexing how retro-or prospective the (B) part is relative to  18 Indeed, if the prosodic-phonetic disjunction before or after the pivot element were any greater for the cases toward the right end of our cline, they would be perceivable as discrete TCUs. the (A) part may in fact prove to be a more omni-relevant concern for participants. Importantly, interrogating these hypotheses will also provide increased opportunities to complement the (primarily) correlational argument that we have developed here, with a discussion of some of the normative and reflexive dimensions involved in participants' uses of, and orientations to, pivot constructions in the service of action (see also Heritage et al. 2019). After all, it is at least theoretically conceivable that speakers can prosodically through-produce pivot utterances in which the post-pivot's action works to occasion a response for the prepivot's action and, vice versa, that speakers can produce pivotal turn-extensions in which the B-part works to re-design the A-part's action without the two parts being tightly integrated. Whether such things actually happen in the real world, and if so, how much they differentially shape the understanding of action (e.g., whether they invite a search for action-level (dis)continuities that might not otherwise have been apparent or whether they are constitutive of further distinct types of action relationships) are empirical questions. Exploration of such cases will also undoubtedly further illuminate the role of the pivot element's syntactic optionality vs. obligatoriness in the constitution of these different action relationships, and especially how those relationships interact with prosodic-phonetic packaging in pivoting as a turn-constructional practice.
We conclude by noting that our findings also raise questions with regard to the way(s) in which actions are conceptualized in research on social interaction (see inter alia Levinson 2013;Schegloff 1996;Sidnell and Enfield 2014;Enfield and Sidnell 2017;Sidnell 2017). As previously described, we started out by focusing on pivot construction candidates that cluster around a single gross action type ("assessments"). In moving forward with our analysis of what speakers seek to accomplish with these turns, it appeared usefulif not necessaryto adopt a more granular perspective on actions-in-interaction. Specifically, it has proven analytically rewarding to zoom in on how the various parts of a pivot construction candidate and the actions they can be seen to deliver relate to each other. This revealed that different kinds of actionrelationships can underpin the use of "pivoting" as a turn-constructional practice. Some of these actionrelationships appear to be symmetric/non-hierarchical in character (as when speakers redesign the same action in a different way), whereas others appear to be more asymmetric/hierarchical in nature (as when a speaker elaborates an action with a "sequentially parasitic", subsidiary action). If actions can indeed be shown to exhibit these different kinds of relationships, then we need a conceptual and terminological apparatus that enables us to capture them more adequately (e.g., to distinguish between independent, main actions and subsidiary, elaborative actions). Moreover, if these different kinds of action-relationships have a bearing on, and are thus reflected in, subtle aspects of turn design (such as their prosodic-phonetic packaging), then the cesura approach adopted herein offering a perspective that avoids treating talk as coming in neatly packaged discrete "units"provides a valuable tool, or an entry point, for unpacking them. It is in this sense that we believe cesuring can inform action analysis, offering new insights into how participants manage the formation and ascription of action in the immediacy of moment-by-moment talk.