A half-century ago, at the close of the pioneering 1962 conference on “Paralinguistics and kinesics” at Indiana University in Bloomington, Margaret Mead grasped for a word that could sum up the conference's aims. That word was semiotics: “[W]e are, I think, conceivably working in a field which in time will include the study of all patterned communication in all modalities” (Sebeok et al. Reference Sebeok, Alfred, Bateson, Folklore and Linguistics1964:275). In time, quite a few of us have, indeed, moved toward modalities other than language, though not at the pace anticipated by Mead, and not always under the banner of semiotics. Still, if Enfield's book is a signpost, the 1962 participants—which included Ray Birdwhistell, Erving Goffman, Edward Hall, and Thomas Sebeok—would be pleased at how far we've come. Drawing on extensive fieldwork and a vast store of Lao video data, Enfield explores speech-with-gesture composites, and the lucid, methodical manner of his investigation—whose phenomena range from lip-pointing to exquisite hand-gestures that depict woven conical fish traps and the logic of sibling-exchange marriages—makes Anatomy of meaning an exciting work. Exciting not just for the way it ventures far beyond linguistics' blinkered sentence-sized units, but for the care it takes en route. It's one thing to invoke multi- or cross-modal relations. It's quite another to show precisely how these relations manifest themselves in communicative events—and without losing oneself, or one's readers, along the way. (See Birdwhistell's (Reference Birdwhistell1970) baroque transcripts in Kinesics and context for a reminder of what can happen when a descriptivist impulse goes unchecked.) Anatomy of meaning is, simply put, in the vanguard of the science of communication. It deserves wide attention.

Those who have spent time poring over video recordings of face-to-face interaction will find themselves nodding in agreement when Enfield rails against the logocentrism that has cordoned off language as an autonomous object, as he does in his opening salvo: “The payoff of studying speech-with-gesture is not only to understand gesture as a phenomenon of interest in itself, but—of greater consequence—to help shake linguists and other students of meaning from the view that language is an encapsulated system for conveying meaning” (ix). “It isn't,” he answers flatly. This, together with his indictment of (neo)-Saussurian approaches to meaning, which privilege the “static, arbitrary, and abstract” rather than the “dynamic, motivated, and concrete” (2), is still news for some, but for fields like linguistic anthropology and sociolinguistics these are pedestrian claims—commonplaces, really—for these traditions have long taken as their foil the Saussurian legacy and its uprooting of language from the matrix of society, culture, and history. In fact, in Enfield's words one hears echoes of Vološinov's (Reference Vološinov, Matejka and Titunik1986) prescient critique of Saussurian structuralism from the 1920s, though readers will find no mention of that genealogy here.

***

If this isn't news, what is? The book's significance lies in its notion of composites—“context-situated composites of multiple signs, part conventional, part non-conventional” (6). Enfield presents this notion in his introduction, then copiously illustrates it in six main chapters that fall into two symmetrical parts. These two parts are defined by semiotic function rather than modality—no accident, for Enfield argues (with Okrent Reference Okrent, Meier, Cormier and Quinto-Pozos2002) that we need a modality-independent account of gesture rather than an account that just presumes modalities are different; to this end he also draws inspiration from Peircian semiotics. (Peirce did not have Saussure's preoccupation with language; his more expansive semiotics has been used to explain everything from honeybee communication to human speech.) Part 1 discusses indexical functions and offers chapters on linguistic deixis, lip-pointing, and hand-pointing, respectively. Iconic, “illustrative” functions command attention in Part 2, with chapters on cross-modal acts of modeling, diagramming, and editing. In each, Enfield treats us to close empirical investigations of Lao video data, which he presents with characteristic lucidity and care. The first-rate quality of the analysis, aided by a judicious use of video frame grabs, figures, and uncluttered transcript segments, makes these chapters not just well argued but admirably accessible.

Still, the book's signature contribution is also its most unsettling question, for it is not always easy to say what composites are, well, made of. Enfield is trying to address a major question here, that of cross-modal integration. For starters, Enfield recounts a remark by Kendon, who suggested that speech and gesture work reciprocally: “the gestural component and the spoken component interact with one another to create a precise and vivid understanding” (Kendon Reference Kendon2004:9, cited in Enfield Reference Enfield2009:9; cf. Agha Reference Agha1996). Enfield rightly asks what “interact” means. Mere spatio-temporal co-occurrence—the fact that speech and gesture co-occur? With his avowed neo-Gricean leanings, Enfield answers this by supplying heuristics—tacit, pragmatic rules-of-thumb that people use to make sense of signs. In this he is not frugal. While presumptive meanings have been neglected in fields like linguistic anthropology (Levinson Reference Levinson2000), Enfield piles heuristic upon heuristic, making the problematic of cross-modal integration feel somewhat cluttered—as cluttered as John Austin's (Reference Austin and Urmson1962) felicity conditions in his classic lectures on performativity. (Remember how it felt like the whole universe had to be analytically reconstructed, from sincerity conditions to procedure to personnel, in order for a metonym like I hereby bet you sixpence—an “illocutionary one-liner,” as Silverstein (Reference Silverstein and Lucy1992:51) wryly put it—to be felicitous, to count as an act?) Enfield posits five heuristics, though we can always add more. For example, there's the orientation heuristic: “if a signer is bodily oriented toward you, most obviously by body position and eye gaze, assume they are addressing you.” There's the unified utterance-meaning heuristic: “assume that contextually associated signs point to a unified, single, addressed utterance-meaning.” There's an agencyheuristic: “if a sign has greater control over a behavior, assume [all things being equal] that this sign is more likely to have been communicatively intended” (17).

At this point we may wonder whether Enfield sees heuristics as the epoxy that holds composites together. When he speaks of a “general theory of composite meaning,” he means “signs co-occurring with other signs, acquiring unified meaning through being interpreted as co-relevant parts of a single whole” (7; emphasis mine). Does Enfield, with Engle (Reference Engle, Gernsbacher and Derry1998), whom he cites favorably, thus want to argue that multi-modal composites are composites by virtue of heuristics that presume integration?

Indeed, this is what Enfield often suggests, but surely we should distinguish, minimally, between (a) presumptive cross-modal integration, where integration is a function of heuristics used by interactants to construe signs; and (b) empirically demonstrable cross-modal integration, that is, the gradient and fractional manner in which co-occurring semiotic resources can be shown to articulate with one another in some event of communication. How interactants sort through and stitch together signs is a vital topic, but this should not come at the expense of a heuristic-independent account of cross-modal integration. It's not that Enfield utterly neglects the latter, because he does, for instance, argue that people rely on “triggers”—signs that guide participants in their effort to draw together otherwise disparate signs. As I address below, at other times he even seems to insinuate that composites stick together through their intrinsic compositionality.

On balance, though, Enfield relies on heuristics to explain cross-modal integration, and this has its costs. After all, researchers have found all sorts of cross-modal relations between speech and gesture, like compensatory relations, where gesture adds information critical to, say, successful reference—information not conveyed by co-occurring speech; or additive relations where one modality contributes increased resolution, as it were; configurational relations where the gesture-speech composite conveys meaning globally and synthetically, and hence cannot be resolved into its parts and their combination. And so on.

It is telling that while Enfield says that composites involve the interplay of conventional and nonconventional sign-vehiclesFootnote ¹ —a fair point—we never really learn what specific empirical forms this interplay takes. On this issue Agha's (Reference Agha2007) discussions of “text” and “textuality” (cf. Hanks Reference Hanks1989; Bauman & Briggs Reference Bauman and Briggs1990; Silverstein & Urban Reference Silverstein and Urban1996; Silverstein Reference Silverstein and Sawyer1997) are instructive. Like Enfield, Agha offers a well-defined place for the categorial values of linguistic form-types, such as deictic expressions, honorifics, and lexicogrammatical resources. Agha demotes these values by calling attention to their precarious status in discourse as “text-defaults,” “in the sense that the item regularly conveys a set of effects in the absence of interference from co-textual effects” (Agha Reference Agha2007:39). Linguistic forms so conceived supply participants with default values, values that are modulated by emergent effects projected by the signs with which they co-occur as elements of their co(n)text, forming a “text-configuration.” Semiotic effects produced concurrently by distinct functional principles (that is, as part of text-configurations) can “illuminate each other during usage,” and “are reciprocally reflexive when considered in relation to each other” (Agha Reference Agha2007:470). Agha appreciates the place of the conventional while detailing a few key ways that a text-configuration can diverge from its text-partials. A text-configuration featuring deictic forms, for instance, may:

A simple illustration of the broader principle:

While Agha does not focus on gesture-speech composites, his sensitivity to the interaction effects of co-occurring signs is critical for another reason. When we have multiple semiotic resources in play we inevitably encounter what Agha calls “fractional congruence”—and that is the stuff of tropes. Consider, for instance, Ochs, Gonzales, & Jacoby (Reference Ochs, Gonzales, Jacoby, Ochs, Schegloff and Thompson1996), who detail how the physicists in their study alternated between physicist-centric constructions like And then I lower the temperature and constructions like If I go below in temperature, where the scientists seem to adopt the perspective of the phenomena under investigation. The patterned deviation from expected co-occurrence patterns helps motivate “transpositional” interpretations—as if the zero-point of deictic reckoning shifts from observer to observed (cf. Hanks Reference Hanks1990; Haviland Reference Haviland, Gumperz and Levinson1996; Perrino Reference Perrino2005). Or consider the familiar cases of tropic departures from second-person singular address forms. (Plural vous for singular addressee in French, a number trope in which one refers to addressee as if s/he were ‘greater than one’; or, in Italian, the displacement to third-person singular feminine Lei (Silverstein Reference Silverstein2003:208). Denotationally “incorrect” reference—again, a matter of fractional congruence—motivates tropic readings of social relations (Agha Reference Agha2007). It is not that gesture-speech “mismatches” (cf. McNeill Reference McNeill1992) need attention (they are apparently quite rare), but that composites generally—especially ritualized ones—can involve juxtapositions of seemingly incompatible elements and thus flout conventions of “unity” for pragmatic effect; we risk missing this when we focus on heuristics and privilege cross-modal unification.

***

Anatomy of meaning may rely too much on heuristics to address the question of cross-modal integration, but on a few occasions it seems to suggest that there's something about composites themselves that makes their signs stick together. Composites, in a word, can exhibit compositionality.

Let us be clear about what this argument is not. This is not an attempt to scramble back to the rarefied heights of high structuralism. This is not, say, Kenneth Pike's (Reference Pike1967) Language in relation to a unified theory of the structure of human behavior, where all sorts of behavioral events were resolved into constituents. In Pike's analysis of a church sermon, for instance, he raced from “articulatory movements” to “sounds” to “words" to “lines” to “stanzas” to the “singing of a hymn” (Pike Reference Pike1967:78)—as if all were amenable to distributional analysis and subject to constraints within syntagmatic “slots.” He culled metadiscursive words and expressions, organized them into a partonymy, then assumed that lexicosemantic orderliness offered a direct window onto the sermon's cultural constituency.

Enfield does not repeat such an exercise, but his “anatomy” trope, trumpeted in the title and echoed by his notion of the composite utterance, does lure us into thinking that there is a specific orderliness that makes composites cohere. Enfield knows perfectly well that some composites are not, strictly speaking, “compositional”—as in sense-compositionality in language; rather, some are configurational and global-synthetic (McNeill Reference McNeill1992, Reference McNeill2005), as in holophrastic lexations like kick the bucket (‘die’), where one cannot predict the meaning of the whole by studying the meaning of the parts and their combination (Agha Reference Agha2007). Still, he is at pains to demonstrate how some gestural sequences can exhibit qualities of compositionality familiar in language. If one were to ask “whether non-linguistic components of composite utterances show linear segmentation and combinatoric dependencies” (113), the answer is yes—they can. In cases of “symmetry-dominance” forms of gestural modeling, detailed in Ch. 5, interactants first hold hands in one position, then one hand does something new against that background. There's nothing surprising about the appearance of something like segmentation and combinatorics in cospeech gesture, but we should take care not to read these qualities into all co-speech gesture nor collapse distinctions among kinds of composites. At times Enfield does exercise care. About the symmetry-dominance gestures, for instance, he adds: “The extent to which this is equivalent to a grammatical construction in speech is open to discussion. It has some of the elements” (148). Linguistic anthropologists have often shown how ritualized discourse tends to exploit the “poetic” function (Jakobson Reference Jakobson and Sebeok1960; Silverstein Reference Silverstein2004). Rituals tend to have a high degree of cross-modal integration, so that they seem segmentable into constituents, like a whole in which subsidiary moves and phases subtend. (Compare with Conversation Analysis, where one adjacency pair can seem nested within another, higher-ranking adjacency pair. See Schegloff Reference Schegloff2007; Sidnell Reference Sidnell2010.) This emergent text-structure exhibits a kind of virtual compositionality that depends on indexical-iconicity (resemblance across co-occurring chunks of text), but this doesn't mean it's the same as constituency in grammar.

In the final analysis, let us read Enfield's composites not as an argument about any one kind of cross-modal orderliness—even though he sometimes invites this reading through his choice of tropes and through showcasing instances of gestural compositionality. Let us read it methodologically, where it is most provocative. Like many uses of “discourse” and “text,” the term “composite” targets, interdiscursively, the autonomization of language as denotational code and its blinkered sentence-size units. If we leave “composite” as an empirical placeholder for different forms of cross-modal orderliness—and disorderliness—then it is free to do what it does best, and that is to perform a methodological displacement (cf. Latour Reference Latour2005), reminding us where the action isn't. It pushes us to see communicative phenomena as Enfield himself sees them—expansively, as multiplex, cross-modal assemblages—and this alone makes Anatomy of meaning an instructive, and indispensable, work.