“M. Marcel Proust has just published the third volume of his childhood memories,” wrote an exceptionally obtuse reviewer of Le Côté de Guermantes in 1920 (qtd. 130). That's one of the generic frames within which À la recherche du temps perdu has been read; other descriptions include “a psychological novel, a novel about time and memory, later a novel about art and aesthetic experience, a philosophical novel, a phenomenological one, and perhaps only later still a novel with a deeply anthropological and sociological bent, a novel interested in how culture happens, how language is a space in which culture happens, in which it is possible to watch culture happening over time, and to watch social change occur as culture happens” (125). The latter is of course the description closest to Michael Lucey's heart; his Proust is an ethnographer of the work done by language, and more generally a kind of wild sociologist who is deeply attentive to the playing out of social roles in conversation and in the idiosyncrasies of speech. The project of What Proust Heard is at once to analyze that narrative mode and to inculcate in the reader a novelistic form of attention to talk and to the social patterns that are revealed and enacted in talk.Two metadiscourses meet in this book. One is the series of reflections made by Proust's narrator on the conversations and the speech patterns of characters in the novel, including (by implication) those of the hero himself. The other is a set of theoretical languages elaborated by Mikhail Bakhtin, Erving Goffman, Pierre Bourdieu, and above all by the linguistic anthropologist Michael Silverstein.Let me start with the former. What the narrator notices and returns to again and again is a depth that underlies all talk (what Nathalie Sarraute calls the sous-conversation) (220). His interest is not in what is said but in how it is said and what that underlying subtext tells us about the speakers. Every act of speech carries with it the traces of a complex personal and interpersonal history, a linguistic habitus displaying the patterns of class, gender, and culture that are mobilized when characters speak. Everyday talk involves the attempt to decipher those traces and to ratify the social roles performed by speakers, as well as the social capital they claim and that is either granted or denied by their interlocutor; and much of the novel's comic force arises from its representation of the failed assumption of a shared habitus.Take, for example, a conversation centered on the Duc de Guermantes that Lucey analyzes at some length. The duke has expressed his regret that his nephew, Robert de Saint-Loup, has aligned himself with the supporters of Dreyfus: “With a name like ‘the Marquis de Saint-Loup,’” he remarks, “one isn't a Dreyfusard.” The narrator then appeals to one of two “laws of language” to explain the duke's use of the inappropriate phrase “with a name like” (“quand on s'appelle”). One of them “demands that we should express ourselves like others in our mental category and not of our caste”; the duke is speaking like a grocer. The other law is that from time to time there arise, as though by pure chance, “modes of expression that one hears in the same decade on the lips of people who have in no way concerted their efforts to use them” (qtd. 16). The first law, as Lucey paraphrases, “has to do with registers and how they are disseminated, and the second has to do, perhaps, with lexical innovation, borrowing, and diffusion” (17). The play of shifting registers continues in this conversation with interventions by an archivist and a historian: The former introduces a lexical novelty (the term “mentality” to describe a state of mind). The latter, “anxious to be part of the conversation,” points out that the word “mentality” occurs more frequently than the adjective “talentuous” (which the duke has just remarked on), citing as his authority the fact that “I'm on a committee at the Ministry of Education where I've heard it on several occasions, and also at my club, the Volney, and even at dinner at M. Émile Ollivier's.” To which the duke scathingly replies—“with false humility but with a vanity so deep-seated that his lips could not refrain from a smile”—that, since he doesn't have the honor of belonging to the Ministry of Education, nor of belonging to the Volney Club (“my only clubs are the Union and the Jockey—I don't believe you belong to the Jockey, monsieur?”), nor of having been invited to dine with M. Ollivier, he hasn't come across the word “mentality” (qtd. 20–21). Social caste is asserted, a social hierarchy reaffirmed.Although characters thus confirm their social roles by repeatedly performing them, their speech is not reducible to the expression of social position: the duke will, under certain circumstances, speak like a grocer. Much of the narrator's metadiscourse is directed to the slippery negotiations of position that take place in talk. Sexual codes are a particularly rich source of positional ambiguity: Charlus has to work out whether being called “one of us” by M. Verdurin means that his sexual proclivities (and thus, surprisingly, Verdurin's) have been revealed or that something more innocuous is intended, and the narrator is driven to anguish by Albertine's abruptly expressed desire to “me faire casser...,” which he reconstructs (“me faire casser le pot”) as referring to an obscene sexual act (158). Aesthetic codes similarly work both to shape a shared social space and to divide it through operations of distinction and exclusion: Elstir, Bergotte, and Vinteuil act as touchstones by which characters judge and are judged (Charlus signals reverence during a performance of the Vinteuil Septet, Mme. Verdurin performs absorption, and the narrator, anticipating Bourdieu, constructs and reconstructs the social field across which judgments of taste and the concomitant judgments of character are distributed).In a series of interludes between chapters Lucey extends Proust's narrator's fascination with the forms and uses of talk to other novelists who share that fascination: Honoré de Balzac, George Eliot, Virginia Woolf, Nathalie Sarraute, and Rachel Cusk. These interludes shift the focus from Proust to the description of a specific genre of the novel that culminates, paradoxically, in Cusk's tonally flat representations of talk which leave it entirely to the reader to make sense of motive and intention: her metadiscourse is present only as inference. We might wonder, though: Which novels would be excluded from this tradition? Which novelists are not particularly interested in the social dynamics of talk? And when Lucey speculates on how a reader would decipher motive and social patterning in a novel made up entirely of dialogue (Ivy Compton-Burnett comes to mind), he doesn't push this to think about how such deciphering might work in the limit case of drama and dramatic irony.The second metadiscourse elaborated in What Proust Heard is drawn from Bakhtin's account of speech genres (but not, unfortunately, except in passing, the early work of Bakhtin/Vološinov and Bakhtin/Medvedev); Goffman's account of the splitting of the author function between animator, author, and principal; Bourdieu's account of habitus, field, and the logic of practice; and Michael Silverstein's linguistic anthropology (but not, unfortunately, the social semiotics of Michael Halliday, which covers much of the same ground as Silverstein but more lucidly and with somewhat more linguistic complexity in its theorization of the interlocking determinations of speech registers). Here the key construct is the “social indexicality” of language, a concept that invites us to attend not to the denotational force of speech but to the social markers it carries and the social work it performs. Lucey's reading of Silverstein is grounded in a compelling critique of the limitations of speech act theory and of H. Paul Grice's “cooperative principle.” The tradition that stretches from J. L. Austin through Grice and John Searle understands language-in-use in ways that are instrumentalist and acontextual, oriented to the speaker's intention and to the truth-values carried by speech. As Goffman puts it, however, talk is less about giving information than about “giving shows” (98). For Silverstein, language has an indexical rather than an illocutionary force: it carries embedded information about role relationships and it projects “consequential social action”; its “indexical entailments” are “more various than normal construals of performativity allow for” (223–24). Whereas Grice's account of conversational implicatures expresses a kind of philosopher's fantasy about the use of talk for the rational and cooperative construction of a shared project, where participants are expected to speak transparently, economically, truthfully, and to the point, “neither Proust nor linguistic anthropology proceeds from the assumption of reciprocity, and so neither finds it normal that an easily or widely shared reference to a common repository of illocutionary possibilities or metapragmatic functions necessarily produces a uniform recognition of what transpires in this or that instance of language use” (32–33). To the contrary: what Proust takes from Dostoevsky is a thematics of mutual incomprehension, such that any conversation between, say, the narrator and Albertine is largely opaque to their respective “intentions,” and speakers are never fully the subjects or authors of their speech. Many of the most interesting and most subtle readings in the book are explorations of the tormented dynamic that holds the narrator and Albertine together and apart.The two metadiscourses elaborated in this book—that of the narrator, and a social-scientific discourse—are interrelated; but the form of their relationship is something of a problem. At times Proust (or the narrator) figures as a kind of precursor of social theory; at others that theory acts as a supplement, translating the narrator's analyses of talk into explicitly sociological terms. I'm not entirely convinced that that act of translation adds significant value, however, and I have a particular concern about Silverstein's concept of the social “indexicality” of language. The Peircean indexical sign operates by means of a direct existential or physical connection between the sign and its object: the connection may be causal, or it may be spatiotemporal, as in the case of deictic indices. The concept of social indexicality thus seems to indicate that role relationships can be read off from spoken text, with immediate legibility, and it points to actual relations of power and solidarity conveyed in talk rather than to relations that are projected but not necessarily accomplished: to discursive positions that don't simply or immediately reflect actual social positions (the duke speaking like a grocer). To put this differently, the logic of the concept of indexicality cuts short an analysis of the negotiation through talk of rhetorical positions, and the interpretive ploys that all such negotiation involves.One of the consequences of this logic is played out in Lucey's response to a reflection by Cusk on the instability of character in her novels. She tells her interviewer that “we're very used to a novel being carried by a self that we believe in, and we believe in it as a simulation, or a representation of us. We think that's what our experiences are like, and this is the form which that experience of being in a self is like, and I don't believe that. I don't think that's true. I think... that experience is much more lateral and oceanic” (258). To which Lucey replies that “of course, it is probably statistically more likely that certain kinds of people will share certain experiences than others, will be informed by certain kinds of experience than others, will talk in one way rather than another.... Cusk's novels constantly hover on the brink of this kind of self-reflexivity. Proust's narrator hovers on this same brink” (259).I don't want to make too much of this: Lucey's reading of Proust is fully attentive to the indeterminacies and instabilities of character and to the “distributive authorship” of speech (276). But one of the ways in which Proust is not doing the kind of norm-governed classificatory work that a sociologist like Bourdieu does has to do with his fascination with the dispersal of “character”: the atavism that finds us speaking the language of our parents and acquiring the racial features of our ancestors; the slippage of one character into another through the myriad transformations of desire and identity in this most fluid of novels. Proust may indeed have an ethnographic sensibility, but his commitment as a novelist is not only to the descriptive but also to the transformative and the Cratylic powers of language.