The Interdependence of Bodily Demonstrations and Clausal Syntax

Abstract

Units in interaction are emergent real-time phenomena that can be accomplished by the coordinated deployment of language and the body. Focusing mostly on data from dance classes, this study looks at how incomplete syntax projects a continuation realized by the body and systematically accounts for clausal syntax that can incorporate an embodied demonstration. It is argued that the classic list of types of turn constructional units by Sacks et al. (1974) needs to be expanded with a syntactic-bodily one and that the syntax of embodied demonstrations has to be included in the grammatical description of language.

Units produced in spoken interaction and written text are not entirely compatible. Traditional grammars have focused on the latter and cared much less about the units of spoken language. There has been a written-language bias in much of linguistics (Linell, 2005). Attempts to deal with units in spoken language have led us to realize the crucial importance of intonation (Chafe, 1993; DuBois, 1991; Ford & Thompson, 1996; Schuetze-Coburn, Shapley, & Weber, 1992; Selting, 1996) and the emergent nature of grammar (Helasvuo, 2001a; Hopper, 1998) that among other things allows renegotiation of unit completion (Auer, 1992, 2009; Couper-Kuhlen & Ono, 2007). Speakers project and complete units in real time, sometimes in collaboration with other participants.

One outcome of the written-language bias in the analysis of language is that descriptions of language have resided with different versions of “ellipsis” in order to explain the phenomena that do not conform with the well-formedness criteria of syntactic units in written texts. In fact, speakers may produce units that include fewer items than a complete clause or sentence would and still make sense in the specific context. Several recent studies have demonstrated how syntactic structures arise as situated actions and argued that grammar is intimately connected to the kinds of actions it recurrently accomplishes (Couper-Kuhlen & Thompson, 2005a, 2008). This article brings together the traditional linguistic concern for the typology of syntactic units and the embodied behavior of participants in interactional settings, focusing on units that employ “elliptic” syntax as well as body movements. It shows that units are essentially emergent in participants’ real-time actions but also rely on the existence of atemporal abstract categories formed over exemplars of language use (Couper-Kuhlen & Thompson, 2011), such as clauses. Originally, units in a conversation were described as follows (Sacks, Schegloff, & Jefferson, 1974, p. 702).

There are various unit-types with which a speaker may set out to construct a turn. Unit-types for English include sentential, clausal, phrasal, and lexical constructions. Instances of the unit-types so usable allow a projection of the unit-type under way, and what, roughly, it will take for an instance of that unit-type to be completed.

As will be shown, the emerging units may also be multimodal, thus not resulting in syntactic or intonational unity but instead being composite structures accomplished by finely coordinated linguistic and bodily behavior.

Two crucial features of a unit are projectability in time and completion, as recognizable for the participants. Projection means that an individual action or part of it foreshadows another one in the near future (Auer, 2005; Streeck & Jordan, 2009), such as an article projects a noun and a story preface projects a story. Projectability seems to be an inherent characteristic of the linguistic-vocal mode of human action, while the realization of the projection can be embodied, a phenomenon called embodied completion by Olsher (2004). Thus, units are not always accomplished exclusively within the vocal channel, which forces us to rethink the nature of units and to expand the previous list of unit-types by Sacks et al. (1974). Human beings constitute meaning and action in a way that integrates speech and the body (Enfield, 2009; Goodwin, 2000, 2002; Jones & LeBaron, 2002; Schegloff, 2005; Sidnell, 2006; Stivers & Sidnell, 2005). Meanings can, for example, be achieved by combining the verbal mode of communication with various kinds of gesture, and the gestures may contain absolutely crucial information. Enfield (2009) has called the resulting units composite utterances, Slama-Cazacu (1976) talks about mixed syntax, and Goodwin's (2006) term is multimodal utterance. However, since uttering involves speech, the term utterance may not be the best one to characterize instances when speech is not clearly profiled. It may convey the assumption that speech is somehow still most crucial for communication. Reference to syntax involves the same limitations. Slama-Cazacu (1976, pp. 216–222) talks about kinesic elements “replacing” syntactic elements, thereby treating language as primary in relation to other semiotic devices. A neutral term in this sense would be hybrid construction, coined by Mori and Hayashi (2006) for characterizing interactional moves involving both talk and gesture in conversation between L1 and L2 speakers, but construction is also a differently defined term in contemporary linguistics. The current study therefore uses the term syntactic-bodily unit, building on the conversation analytic tradition (especially the programmatic article by Sacks et al., 1974) and being transparent about what the structure consists of.

Units emerge in time: A speaker initiates a contribution, projects its continuation throughout the emerging structure, and foreshadows or even accomplishes its completion. Syntax of a language plays a crucial role in this endeavor and thus deserves special attention. In fact, it has been argued that one of the most important jobs of syntax in the organization of talk-in-interaction is to provide potential construction and recognition guides for the realization of the possible completion points of turn construction units (Schegloff, 1996, p. 87). This article characterizes the projective syntax of multimodal units emerging in dance instruction and the realization of the projection in embodied completions. It concentrates on cases where obligatory elements of clauses are “missing,” as this is where grammatical projection is nonnegotiable, such as when an argument of a transitive verb is not yet uttered. Furthermore, clauses as syntactic units are clearly oriented to by participants in interaction, as has been shown for the typologically different Finnish (Helasvuo, 2001b) and English (Couper-Kuhlen & Thompson, 2005b). In contrast with earlier studies on multimodal constructions, the current article does not include cases where a deictic item in talk relates to an embodied performance or gesture, as this would make the unit syntactically complete. Furthermore, this would involve a different kind of mutual timing of the language and the body in comparison with cases when a clause lacks a grammatically obligatory element.

Grammar and body movements are complementary devices of sense making for human beings, and sometimes they cannot be analyzed separately without losing the essence of what is conveyed. General awareness of bodily involvement in human interaction has risen together with the availability of video recording that enables researchers to repeatedly scrutinize visual information. First video-based microscale interaction analysis was put forward about three decades ago and concerned gaze (Goodwin, 1980). Gestures have been widely analyzed, not only experimentally but also in terms of turn taking and storytelling (e.g., Kendon, 1983; Sidnell, 2006; Streeck, 2002, 2009). Pointing in particular has been in focus in the studies that consider both language and the body as meaningful devices in interaction (Goodwin, 1996, 2003; Hayashi, 2005; Hindmarsh & Heath, 2000; Mondada, 2007). Building on the methodology developed in these studies, the current one looks at an activity that concerns the whole body. So far, mostly body orientation has been considered and how it relates to the opportunities of participation (e.g., Goodwin, 2000; Hindmarsh & Heath, 2000; Schegloff, 1998). In this study, the body is actively moving in order to make sense. Furthermore, topics that have drawn predominant research interest in the interactional studies—pointing, gesturing, and gaze—are of limited use in dance classes, as they can distort the movement being demonstrated. Whole-body demonstrations in dance classes will thus constitute a qualitative complement to the previously mentioned studies.

The article is primarily based on dance class data and describes the regular incorporation of embodied demonstrations into clauses. The basic pattern will be presented first, moving on to the systematic account of clausal structures that enable embodied elements. Further, different types of embodied demonstrations will be discussed in regard to their timing with the syntax in progress. Finally, the specifics of projection and completion in embodied behavior are discussed insofar as they have a bearing on the basic pattern of syntactic-bodily units: syntax first, embodiment later.

THE DATA

The data were video recorded in classes of Lindy Hop, Balboa, step, and ballet in 2006–2011, consisting of about 38 hours, and the 17 teachers speak English, Swedish (Sw.), or Estonian (Est.). The recordings include beginner through advanced levels of nonprofessional dancing. Lindy Hop and Balboa are partner dances, and in the recorded classes they are taught as a social dance, not as a choreographed performance dance. Therefore, the instruction is often about leading and following within the dancers’ couple. The corresponding dancer roles are called the lead(er) and the follow(er), which will be referred to in the examples. Accordingly, the teachers are called Lead as in lead teacher and Follow as in follow teacher. These classes are ordinarily taught by two teachers who can both address problems with any dancer role. Ballet and step classes are taught by a single teacher, and the dances are choreographed.

On the basis of these data, a collection of about 220 cases of clausal syntax incorporating bodily demonstrations has been compiled. The qualitative analysis is supported by the regularities present in this large collection. Additional instances from alternative activities, such as language teaching, master classes in music, physiotherapy, and L2 conversation, have been collected from published sources. A dance class, however, is an exceptionally prolific setting for finding out the various options of incorporating embodied demonstrations into syntax.

THE BASIC PATTERN

For a start, let us look at a typical example of the pattern. The teacher is explaining a problem in the students’ dance and suggests a correction. As part of the corrective segment of teaching, he demonstrates what he has seen the students do, as shown in Example (1). Transcription conventions can be found at the end of the article. The embodied demonstrations are only represented by the label “demo” in the transcripts, as the complex drawings of dance moves would necessitate a great deal of insight into the activity and distract the reader from the argument of the article.

There is a syntactic initiation of a unit (marked in bold), which is completed by an embodied demonstration. The initiation projects a continuation, I see guys go is syntactically incomplete, as go is here used in quotative meaning making relevant an object. Also the nonfinal prosody supports the projection, as the lead teacher keeps his pitch level on guys go. The completion of the unit is evidenced by students’ laughter that comes after the teacher has completed his bodily demonstration. Laughter is a relevant response here, since the demonstration has been exaggerated as a caricature (analysis of cases like this as embodied quotes and their pedagogical functions can be found in Keevallik, 2010). Recipient reactions such as laughter witness the end of a unit. During multiunit turns, recipients regularly mark their understanding of the so-called intermittent unit completion by using minimal responses (Selting, 2000, p. 511). That brings us to the issue of defining an interactional unit.

What it takes to accomplish a meaningful action depends on the activity and the precise sequential context of the unit. Some activities, such as storytelling or instruction, generally take more effort and time than, e.g., asking a question or agreeing to help, and accordingly warrant longer multiunit turns. One of the controversial issues is how to describe the smaller segments of these longer turns. It is obvious that extended actions consist of several syntactic, clausal, etc., segments, which are called turn constructional units (TCUs) in the conversation analytic tradition. At the same time, TCUs are defined as potentially complete turns on the completion of which a transition to a next speaker becomes relevant (e.g., Schegloff, 1996, p. 55). It is clear that in storytelling a speaker transition is not relevant after every smaller segment, except for tokens of recipiency and appreciation. Therefore, Selting (2000) suggested that the notions of speaker transition and TCU be kept apart. She proposed her own definition of the TCU along the lines of the linguistic tradition: “the smallest interactionally relevant complete linguistic unit, in a given context, that is constructed with syntactic and prosodic resources within their semantic, pragmatic, activity-type-specific, and sequential conversational context” (p. 477). TCUs as building blocks are thereby separated from the action-dependent matter of speaker transition.

This line of argument is supported by actions by other participants whose displays of recipiency regularly occur at the intermittent completion points during longer turns. In other words, participants in talk orient to the completion of some units even though the larger action is not yet complete. The current study subscribes to the understanding of units as not necessarily making relevant turn transition but as a practice of forming an interpretable segment of verbal-embodied behavior understood to be in some sense complete by the participants. The dance class data involve a turn-taking structure with preallocated longer turns for the teachers, while the students are not in the first hand concerned with immediately taking the next turn. Nevertheless, some kinds of units emerge one by one in teacher talk in real time, each displaying an initiation and often also a completion, which is potentially oriented to by the participants. A unit is indeed an epiphenomenon of turn holding, yielding, starting, and ending, but it is still retrospectively recognizable as a unit (Selting, 2000, p. 491). Analyzing these units implies understanding practices used to build turns, especially longer ones, and meaningful action in context.

In short, in Example (1), we can see a unit consisting of a verbal and a bodily component, which is treated as complete by the participants. Schematically, the temporal pattern of syntactic-bodily units looks like Figure 1.

FIGURE 1 The basic syntactic-bodily unit.

Temporality is crucial in the pattern. It lies in the nature of projection that something that is produced first foreshadows a part that comes later. The absolute majority of the cases in the current database display this temporal structure: first incomplete syntax and then the embodied demonstration. The unit is potentially complete after the embodiment, thus satisfying the main criterion of a turn constructional unit. There are only four cases (out of about 220) where the ordering is different, with the embodiment coming first. These instances will be discussed separately.

It has been clear for a while that relevant information for human interactants is not necessarily exclusively conveyed in the stream of speech. Early examples of this kind of reasoning are Slama-Cazacu (1976) and Kendon (1983). Two studies on language learners and one on language teaching have shown how L2 speakers as well as teachers complete turns with embodiments in order to make sense for their interlocutors (Lerner, 1995; Mori & Hayashi, 2006; Olsher, 2004). It has also been argued that gestures are a crucial device of sense making for an aphasic man who is able to produce only three words (Goodwin, 2006) and for a speaker describing fishing devices and kinship structures in his community (Enfield, 2009). However, nobody has yet systematically looked at clausal syntax that enables incorporation of embodiments. Supposedly, embodiments cannot be inserted just anywhere in the ongoing clause.

This is not to say that embodied demonstrations cannot be accomplished during the production of grammatically full clauses. They can indeed. In the dance class data a verbal description may accompany a bodily demonstration in perfect synchrony, with prosodic prominences falling on accentuated steps or focused moves. There may also be pauses in talk during part of the demonstration, or a demonstration may start somewhat earlier than its verbal gloss. Several kinds of timing are possible, but the focus in the current study is on cases where the body explicitly gets a syntactic role—where the participants treat the embodiments as completing syntactic units that would otherwise have to be perceived as elliptic, incomplete. The patterns described constitute a specific part of human grammar that should be described in its own right and included in comprehensive grammatical descriptions, along with the many constructions that exclusively appear in the written language but have always been a self-evident component of grammars.

A NOTE ON ACCOMPANIMENTS OF EMBODIED DEMONSTRATIONS

Before moving on to the analysis of the pattern, a note on vocal activity during the demonstrations is in order. The embodied demonstrations in dance classes are regularly accompanied by vocalizations, which are not conventional words of the language and thus also not part of the syntactic structure being produced. One type of vocal action that is closely tied to physical performance but does in fact use elements of language is counting rhythm with numbers. Teachers regularly do it to coordinate the timing of the practice or to accompany their own demonstrations. These numbers do not as a rule participate in syntax. A related type of vocal action is the use of various syllables to mark rhythm and simultaneously characterize the movements. These vocalizations are nonlexical and also used in musical training, traditional music and dance, as well as scat singing. Example (2), which is an expansion of (1), displays both types of movement-accompanying vocal action. In line 9, the teacher first counts the rhythm with beat numbers and then produces vocalizations during the demonstration of steps.

For the snippets of projective syntax, it does not matter whether beat numbers or vocalizations accompany the demonstration, or whether the teacher is silent. In Example (2), the syntax is close to identical in lines 5 and 8. Vocal accompaniments thus belong to the embodied demonstration rather than being part of the syntax. There are several reasons for making this analytical distinction: Accompaniments use a very limited lexicon, can be varied in an ad hoc way, and produced in chorus by both teachers and students. In addition, there is basically no morphosyntax involved, while the prosody is distinctively related to dance rhythm and musical bars, displaying large voice modulations and even singing. Besides vocal accompaniments, other audible effects can be used during the performance, such as clapping, snapping, and stomping. These may be implemented interchangeably or simultaneously with the vocal devices, with or without music, and with or without introductory syntax. All of this shows that accompaniments constitute a specific mode of behavior that is not quite on the same level as what we traditionally analyze as syntax and lexicon.

In the following, syntactic-bodily units accomplished with and without accompaniments will be discussed together, unless there are specific limitations in some subpattern. In most cases, the teacher has a choice to be quiet during the demonstration or to produce some accompaniment. However, the latter constitutes a pedagogical advantage, enabling mutual timing of the dance practice, marking bodily accents and even the character of the movements. In this way, the accompaniments are activity specific, but they are also used in other contexts of physical and auditory practice, such as musical rehearsals and sports training. One could expect other kinds of vocal effects to be used in other activities, such as onomatopoetic sounds in adolescent storytelling, with similar syntactic structures. This could be a subject of further inquiry.

THE PROJECTIVE SYNTAX

Let us now look closer at the syntactic elements in the first part of the unit. The projective syntactic structures are boldfaced in the examples, and the placement of the embodied demonstrations is aligned with the line of talk, representing the emergence of the multimodal patterns in actual time.

The syntactic structures often involve an agent, either expressed by a pronoun or a category name (killarna [“guys”], guys, you guys, we). Agency is ascribed with words and can probably not be done as easily by the body. Another characteristic feature of the syntactic part of the units is that the structures involve modal verbs, negation, and conjunctions (such as so, om [Sw.“if”], då [Sw.“then”], siis [Est. “then”])—matters that can easily be expressed with linguistic structures. The same goes for conditionality (if … then) (Example 4); it would be much more cumbersome to express conditionality with the body only. A first conclusion is therefore that the syntax that projects embodied constituents involves elements that are hard to represent by the body. This may be one of the main reasons why language is widely used in dance classes, even though it is essentially a body-focused activity. Some meanings can most efficiently be conveyed by language.

The overwhelmingly most-frequent type of embodiment projection is accomplished by a verb constituting (a part of) the predicate (about 180 cases). That includes instances with copula, such as it's + demo, det blir [Sw. “it'll be”] + demo. In Swedish and Estonian, the dummy subject and the verb appear in the reverse order in case the syntactic unit starts with an adverb, as in sen är det + demo [Sw. “then it is,” lit. “then is it”], or nu har vi + demo [Sw. “now we have,” literally “now have we”]. Thus, the item just before the embodiment in these languages is not the verb but a dummy subject or a pronoun. Another option for the syntactic initiation is to produce only the modal verb of the verb phrase or other auxiliary verb, leaving open the slot where the main verb would go in regular syntax, as in just try to- + demo and så kan jag då [Sw. “then I can”] + demo. There are also quotative items, such as go in English, illustrated in (1) and (2), and the particle ba in Swedish (analyzed as quotative grammaticalized from an adverb by Eriksson, 1995 ¹ ). In dance instruction, the quotatives introduce quoted demonstrations that are attributed to specific agents (Keevallik, 2010). The use of quotative markers may be accounted for within the same category of projective initiations, even though a quotative element does not have to be a verb in all languages.

Thus, the most frequent type of projection that enables embodied demonstration is carried out with the particle/conjunction + agent/dummy subject + main or auxiliary verb/predicate, at least in the three languages studied here. Without further cross-linguistic data, nothing can be assumed about the universality of the pattern, as many languages do not organize their clauses in the order of noun phrase + transitive verb + noun phrase. For example, it would be very interesting to know how a verb-final language accomplishes projection of embodiments, if at all.

When it comes to the choice of verbs projecting the demonstration, one of the salient features is that they tend to be generic. This underscores their subsidiary role in regard to the upcoming bodily demonstration, which constitutes the pragmatic focus of the entire structure. Frequent verbs in the current database include be, have, do, see, go, and come. This is one reason why it would be problematic to consider language the primary communicative mode in a dance class: The fully specified information for dance students is presented by the body and the generic syntactic initiations play a supportive role. The priority of language is a feature of some but not all human activities.

The verb choice is to some extent activity specific. In order to use the verb for seeing, as happens in the data, the activity has to be visually observable, and lead is an action-specific verb that can be used for instruction in social dances that are leadable. In Weeks's (1996) data on an orchestra rehearsal, the verb play is used, which would not make sense in dance instruction. Otherwise the structures reported by Weeks are syntactically identical with the ones in the dance data. Data from several other activities support the generic pattern. Haviland (2007) discusses data from a musical instruction setting, and his examples of incomplete syntax include the transitive verb in you guys are doing + demo, a modal verb in so you can + demo, so that you can + demo, and a copula in so that you have less possibility to be + demo. Martin's (2004) data come from physiotherapy sessions and include a modal auxiliary in å så sku jag [Sw. “and then I should”] + demo. Data from language classes (Lerner, 1995) and second-language conversation (Olsher, 2004) include copulas doorknob is + demo, paste is + demo, and but you are + demo. A mundane activity such as storytelling (Sidnell, 2006) can also display structures such as all ‘fa sud’ (this guy goes) + demo (of a car horn). The initiating syntax + embodiment patterns could thus be explored across activities and settings in order to find the universals. What already becomes clear in the available sources, however, is that the variability of lexical items and grammatical structures in the syntactic-bodily units is too large for it to be analyzed as a linguistic construction.

When it comes to further structures, in addition to the predicate the speaker-performer can also produce an article, preposition, determiner, or other dependent element in the noun phrase to preface the embodiment, such as where we are in + demo, or you finish it off with a little + demo. An example of an article is shown in (6).

The analysis of cases like this could be that the slot that is filled with an embodied demonstration is that of a noun in regular grammar (about 20 cases in the data). Similar instances can be found with slots for adjectives (about five in the data), as shown in (7) but it should be clear that the embodied demonstrations constitute a specific component and do not display nominal or adjectival features (hence the tentative counts). In (7) a dependent intensifier is produced to project a demonstration, a slot that could otherwise be filled with an adjective.

Demonstrations could also be used in the slot for an adverb (about 10 cases):

However, when it comes to elements that are not obligatory in a clause, such as adverbs, the grammatical projection is somewhat less clear. Since the talk before the demonstration could constitute a syntactically complete unit, it is harder to argue that a syntactic element is still projected. Projection is in these cases pragmatic and prosodic rather than purely grammatical. For example, in (8), the speaker clearly does not produce a terminal prosodic contour on we lead, which is therefore heard as projecting a continuation.

Last, in extremely rare cases (a couple in the data) only an agent is expressed and the rest is embodied. In Example (9), the syntactic unit involves a conjunction and the subject she.

The coordination of the syntax and the body is here also more complex, as the pronoun refers to the speaker's partner, and she in fact does the demonstration. Indeed, the speaker as a lead is leading the move and thus organizes the timing so that she can do the demonstration at the right moment. The result is a coproduced unit in which the projective verbal segment is uttered by the lead teacher and the embodied completion is done by the follow. Coconstruction of the syntactic-bodily units will be discussed further in the next section.

In summary, we have seen that an embodied demonstration can be immediately preceded by expressions of agents, dependent elements of noun and adjective phrases, but most often they are projected by (auxiliary) verbs and quotatives. Crucially, just like any projection, the incomplete syntax is not determining or predictive—the unit can still be deserted or simply terminated by verbal means. We can only tell retrospectively which type of unit it turned out to be. The projective syntax provides a scaffold for the embodiment in the form of grammatical structure and lexical framing. Dependent items, agents, and generic verbs are expressed in language, leaving the informatively crucial slots of objects, adjectives, adverbs, main verbs, and quotations potentially open for embodied demonstrations. The embodiments as a rule provide the new and focused information of the unit. The feature that binds all the patterns together is temporality: language first, embodiment later. The projective capacity of grammar is used to accomplish an interactional unit that includes a grammatically obligatory embodied component.

OTHER- AND POSTCOMPLETION

The strongest proof of projection is interpersonal behavior: cases when a speaker's projection is realized by another participant, who thereby demonstrates his understanding that there indeed was a projection. In Example (10), student A has asked a question of the lead teacher, who starts to answer in line 1. He comes to a halt, and the student uses the pause to terminate the unit by showing his own (unreasonable) version of the step. In this case, the verbal and the embodied segments of the pattern are produced by different persons: The teacher syntactically initiates a unit but halts its production, and the student completes it with his own body, showing his understanding of how the teacher must have stepped. Again, we see a laughing reaction by other students once the demonstration is done (line 3).

Other-completion is common within the teaching couple. One of them may comment on the movements they are jointly involved in, as happened earlier in Example (9). Interactionally they form a single party, while in (10) the dynamics are different. The teacher is the expert trying to make sense of the problem the student is asking about. The student embodies it, allowing the teacher to provide a correction, which he does in line 4. In any case, coconstruction of the complex syntactic-bodily unit shows that the incomplete syntax is indeed understood by the participants as projecting a continuation.

Furthermore, interactional units are always negotiable and flexible at the end; they can be recompleted (Auer, 1992, 2009; Couper-Kuhlen & Ono, 2007). Any completion is merely a potential one, and the speaker herself or another participant may choose to add something incrementally. Similarly to verbal-only units, a recompletion may be accomplished in the case when the unit already has been completed by an embodied demonstration. There are two options for this. First, a new syntactic element may be added to the clause. Second, the embodied demonstration may receive a gloss.

In Example (11), the teacher incrementally adds an adverbial phrase after the embodiment in line 3.

The adverbial phrase does not characterize the embodiment that was just demonstrated but adds a piece of information. The unit is thus first potentially complete after the embodied demonstration, and then an increment is added. The term increment refers to the extension of a prior unit that could have been understood as completed. It builds on the syntax and semantics of the prior unit and does not initiate a new one (Couper-Kuhlen & Ono, 2007; Schegloff, 1996). This interpretation is underlined by the very low intonation on the increment in line 3 in Example (11), and its fast production. It recompletes the unit that was potentially already complete with falling intonation on khum, and the increment is therefore produced even lower, which is one possible prosodic characteristic of increments (as discussed by Walker, 2004).

In contrast, a gloss for the embodied demonstration is shown in Example (12). The teacher is illustrating what can happen when the lead dancer does not “sit down” to counterbalance his partner. The teacher initiates the syntactic unit in line 2 and continues with the embodied demonstration, falling forward on the o::::. He then chooses to give a gloss for his movement in line 5.

The unit is first potentially completed by the embodied fall in line 3, and then recompleted with the gloss. Crucially, the bystanding follow teacher already reacts to the embodied demonstration, displaying an understanding that the unit was complete there and that recipient action is an option. In line 4 she nods, looks around at the students and at the same time points at her partner, underlining the importance of his point. The mmm is produced as an emphatic approval of what he is explaining. The verbal gloss comes after that and is added incrementally, building further on the syntax that was initiated in line 2. The end product can be heard as a perfectly grammatical sentence without any ellipsis. However, to ignore what has happened on the way there would be interactionally misleading. In summary, this section of the article argued that multimodal units can be coconstructed as well as recompleted, similarly to units built exclusively with verbal means.

ALTERNATIVE ORDER OF INCORPORATION OF THE EMBODIED DEMONSTRATION

There are four cases in the current database in which the embodied demonstration is incorporated into syntax in a different way, violating the temporal structure “language first, body later.” A case in point is shown in Example (13). The class is working on a longer step sequence piece by piece and has just practiced the first section accompanied by the teacher's di:gida gidi:gida. It has been made clear that there are some steps left after the students have mastered this part. In the excerpt, the teacher once again guides the students through the di:gida gidi:gida but then adds the last segment of the step sequence accompanied by wak:ido. After wak:ido she completes the clause, incorporating the movement during wak:ido as the initially positioned object of the clause in line 3. Without it the clause will be added now (literally “will we add now”) would lack an obligatory element.

Thus, peripherally it is possible to have the embodiment first and the grammar later, apparently involving similarly vague lexicon. In Example (13), this is of course enabled by the common awareness in the class of the overall structure of the step sequence and the fact that they have come to this very point in their practice when the last part of the step sequence can naturally be added. Another factor that enables the particular grammatical structure is the reverse word order of subject and verb in Swedish grammar in object-initial clauses.

It would, however, be difficult to argue that the vocalization wakido or the embodied demonstration itself projects the upcoming syntax. In principle, the embodiment could stand on its own at this particular moment in the dance class, and it would still be understood to be the same kind of instruction. In other words, projection primarily seems to work within the linguistic structure, and a movement (+ accompaniment) cannot quite project in the same way. A body movement does not specify what it will take to complete the unit, as grammar does. A movement is potentially complete on its own; only occasionally a return to home position is relevant. Even though the option with embodiment first and syntax later is an interesting one and also challenges the autonomy of language as a system, it should be underlined that this is a rare structure. In addition, in all of the four deviant cases the demonstrations are done with a vocal accompaniment, which might be a prerequisite for the reverse pattern to work in the first place.

THE EMBODIED DEMONSTRATION

The Timing

The embodied demonstration is as a rule finely timed with the projective syntax so that the focus move starts precisely when the syntax stops. However, an embodied demonstration necessitates a preparation in order for the focus movement(s) to start on the beat. Going back to Example (2), the teacher prepares during both projective segments, as shown in (14). In line 5, he moves his body weight to the right foot (R), lifts the left leg (L), and snaps with the fingers (S) to establish a beat. In lines 7–8, he does two preparatory steps with left (L) and right (R) foot and lifts the left one (UP). The physical preparation contributes to the understanding that an embodied demonstration is about to follow. In this sense, the preparatory moves project an embodied demonstration.

In other cases, the speakers may already be dancing when the projection with grammar is done. Thus, there are two types of timing of the pattern: from standing, as shown in (14), or in the middle of the dance. In Example (11), the teachers are dancing and the lead teacher is talking during the dance. The embodied demonstration with its accompaniment is timed with the talk so that the focused dance segment ends up appearing in the right place within the syntax, as shown in line 2 of the transcript. Similarly to talk, dance movements take real time to produce, and while talk can be compressed, dance beats have a conventional length. ² Thus, it is often the pace of talk that is adjusted to the timing of the body. Pauses and lengthenings may be used to fill the time until the demonstration can begin, or alternatively, talk may be compressed so that the embodied demonstration starts free of it.

In short, the timing of the bodily and vocal behavior is absolutely crucial for the pattern to exist. For an embodiment to receive a syntactic role in an ongoing unit, it has to be on time and appear in next position, just like a corresponding word would. In general, the first step of the demonstration or the focused move comes after the last word of the verbal segment. Physical preparation for the dance, such as taking the dance position with one's partner, bending knees, and lifting a leg, all support the projective force of the syntax in these multimodal syntactic-bodily units and specify the continuation as embodied.

The Character

In contrast to cases where dance movement is presented in its full form by the teacher(s), it can also be decomposed for pedagogical purposes. The teachers can demonstrate leg movements, posture, rhythm, etc., separately. In Example (15), the follow teacher has abandoned his lead and carries out the instruction alone. The dancing couple is thus dismantled, and the comment is about the general manner of dancing. The teacher is standing still on a spot and shows metaphorically with her hand how the dance “dies.” She embodies a hesitating posture, a hunched back, and raised shoulders, illustrating the dying, but she does not actually dance.

Thus, in order to comment on some facet of the performance, the teachers may choose to segment or decompose it so that the demonstration does not look like dance any more. A pedagogically vivid demonstration may also be accomplished by using the body in alternative ways.

The Length (of the Projection)

The length of the demonstration varies, as does its complexity. This has of course to do with the local pedagogical needs but raises an important theoretical question about the projection of the completion within a syntactic-bodily unit. On the basis of available materials, it has not been possible to establish any correlation between the type of syntactic slot (after main verb, auxiliary, article, etc.) and the length or nature of the demonstration. Besides pedagogical needs, the length of the embodied demonstrations in dance classes are defined by dance segments and musical bars, thus not language. Furthermore, intonation on accompaniments can be more informative about projection length than the intonation on the projective syntax. In case the teacher initiates the accompaniment high in his/her pitch range as if in a longer sentence, it can be assumed that the demonstration lasts at least one musical “sentence.” The interactional implications of accompaniment intonation could thus be an interesting research topic on its own.

The demonstrations can be short but also quite extensive. When the demonstration concerns just one move, there might be a single syllable allocated to it in talk, as happened in Example (11). When a longer segment of the dance is presented, the completion of the unit becomes a practical issue for the participants. For example, in Example (5), the teachers dance for a considerable amount of time after the initiating syntax so we have. They orient to the dance timing defined by musical bars and perform multiples of eight counts. The focus step appears in the third one. We can still see student reactions after the crucial step has been performed, but the performance does not stop there, as there are multiple organizations at play at once. On the one hand the teachers are involved in the dance with its internal organization, on the other they participate in a pedagogical setting, offering an answer to a student's question. Accordingly, a dance demonstration may start and end as appropriate for the organization of dance rather than instruction, or the other way round. Its length is not projected by the verbal segment, which is a crucial difference from the beginnings of other types of turn constructional units. Orientation to two different orders furthermore results in a variable timing of recipient action: Students may nod or start a subdued tryout after the focus step without waiting for the entire demonstration to end. Alternatively, if the purpose of the demonstration is to illustrate an incorrect performance as a caricature, recipient laughter may start before the demonstration ends. In short, some aspects of projection length and recipient action are clearly activity dependent. More generally, it seems that a bodily action can project that a completion will come at some point, but not quite the length and details of what it takes to arrive there. This is different from clausal grammar in a language, where projection may be quite specific, sometimes down to the semantic category of the noun (e.g., locative). The completion of an embodied demonstration is clear in retrospect. ³

However, there are also structures in noninstructional conversation that are expandable for a considerable amount of time across several subunits (Auer, 1992; Günthner, 2011), so expandability seems to be a basic feature of human communicative action. Unit initiations such as so we have (5), you guys go (2), and there is the (7) function as story prefaces or projective constructions in talk, such as the German die Sache ist (“the thing is”) (Günthner, 2011). In a parallel way to this kind of constructions, the projective syntax described previously does not show exactly how long it takes for the unit to be completed.

In the current data, the demonstrations can vary from full-length step sequences to decomposed short chunks of the dance. Demonstrations can be done from leisure position or during the dance in progress. For the organization of syntax, it is crucial that the demonstration or its focal point falls temporarily close to the syntactic initiation. The step sequence starts either on the beat after projective syntax or during the already ongoing syntax, with the focus step falling on the beat after initiating syntax. In this way, the emerging units in dance classes can consist of both verbal and embodied elements in an ultimately intertwined manner. Their difference from verbal-only units is that the length of the projection is relatively more indeterminate and rather defined by matters of dance organization than grammar.

CONCLUSION

This article suggests that embodied demonstrations are treated as elements within turn constructional units in dance classes. A speaker may either be quiet or vocalize rhythm during the demonstration. In both ways, she leaves a temporal space for a demonstration, and the demonstration ends up embedded in the syntax produced before and occasionally after it. However, we are only just beginning to understand how body and language are organized in relation to one another. Crucially, they can be used simultaneously, full-clause comments can be uttered during an ongoing bodily demonstration, and gestures can be accomplished during ongoing talk (e.g., Goodwin, 2007; Mondada, 2007; among others). The embodied elements cannot therefore be described on a par with verbal elements. Their functioning is specifically restricted by their embodied nature, allowing for the preparation of the demonstration to start during projective talk. Still, the structures enabling embodied demonstrations could be outlined in a grammatical description of a language. They are not simply incomplete or elliptic occurrences of snippets of syntax but represent systematic patterns of human communicative behavior. Based on the English, Swedish, and Estonian data used for this study, a suggestion for the description could look as follows. Possible moments when the syntax can be discontinued for embodied completion are marked with an asterisk. The schema is somewhat simplified, excluding inversion patterns and obliques. Neither does it show that the units can further be incorporated into complex syntax, as we saw happening with the if … then pattern in Example (4).

subject * >

predicate/copula/quotative/auxiliary verb * >

article/preposition/determiner/attribute/intensifier *

For research on interaction, the contribution of the current study is a demonstration that units can consist of syntax and an embodiment in a manner that is very similar to that of units built in language. They display projection that is realized in a completion, but the two are done in different modalities. Thus, the classic list of unit-types proposed by Sacks et al. (1974), including “sentential, clausal, phrasal, and lexical constructions,” could be expanded with syntactic-bodily constructions. The original proposal explicitly concerned English, and English data have been considered in the current study as well.

The article deliberately did not discuss instances of syntax involving deictic elements that may also project a demonstration and very often do in the data. However, with a deictic the syntactic unit is formally completed, even though it may still semantically and pragmatically project a demonstration. Structures with deictics will have to be studied separately, as the timing of embodied demonstrations is likely to be different there (as shown in the case of the deictic here, Keevallik, in press). Nevertheless, even these units could qualify as syntactic-bodily ones, as could a number of other coordinated arrangements of the body and the grammar. The current study only looked at a subset where a combination of incomplete syntax and an embodied demonstration emerges as a complete unit with a recognizable initiation and completion in time.

The study contributes to the growing body of research into communicative patterns that do not account for language separately from other devices of sense making available for humans. Some of them also consider body as not merely having a subsidiary role in the organization of talk (e.g., Kendon, 2004). In particular, embodied reference (Goodwin, 2007; Hindmash & Heath, 2000) and nodding as an embodied response have been studied (McClave, 2000; Stivers, 2008). This study shows that there are activities where the entire body is involved in the building of communicative units. In dance teaching it is obviously the body that conveys the most elaborate information, while language is only able to accomplish an approximate gloss. This urges us to ask what language is good for. There appear to be matters that are most efficiently communicated in the symbolic form, such as contrast, modality, attributed agency, conditionality, and negation. Language and the body play complementary roles in the construction of meaning in physical instruction. Even though the article focused almost exclusively on the interactional setting of dance classes, other examples from the literature confirmed that the phenomenon is not restricted to a single domain of language use. Similar patterns occur at least in other types of physical instruction but also in qualitatively different activities, such as foreign-language conversation. There is thus a good reason to consider including these patterns in descriptive grammars. The patterns presented previously call to question the common practice of separating verbal behavior from the rest when it comes to analyzing spoken human interaction and syntax in particular.

Acknowledgments

The author is indebted to Elizabeth Couper-Kuhlen, Anna Lindström, and Sandra Thompson for a great discussion of the article at various stages. The research was funded by Riksbankens Jubileumsfond.

Notes

¹It has not been possible to detect projective items that could qualify as grammaticalized quotatives beyond what has already been described for each of the languages. None of the verbs is used with sufficient frequency for that, neither have they undergone semantic bleaching or phonetic assimilation.

²For pedagogical purposes, the length of a dance beat can also be manipulated, mostly by slowing down.

³Action specifically, dance teachers can also just go on dancing after the demonstration, which happens, e.g., in Example (5). The junction where they stop the demonstration and start dancing for the fun of it is very clear.

APPENDIX

Transcription Conventions

Table

CAPital letters	emphasis
−	truncation
[ ]	overlaps, timing of embodied demonstrations
=	latching of units
(0.5)	pause length in tenths of a second
(.)	micropause
((comment))	transcriber's comment
:	lengthening of a sound
>quicker<	quicker talk
.hh	breathing in
boldface	the initiating syntactic structure
.	pitch fall at the end of an intonation unit
?	pitch rise at the end of an intonation unit
,	level pitch at the end of an intonation unit
[space] -	unfinished intonation unit
/	alternative translations

Abbreviations

Table

DEF	definite
FUT	future tense
PAR	untranslatable particle
PL	plural
REF	reflexive
SG	singular