SOURCE: "Character Voices," in Reading Petronius, The Johns Hopkins University Press, 1990, pp. 137-54.

[In the following essay, Slater contends that it is the content and occasion of language more than its form that results in the sense of individual characterizations in the Satyricon.]

Our initial linear experience (in the previous chapters) of Petronius's novel is over. This experience itself has been a fiction: I who write and, most likely, you who read are not first-time readers of the Satyricon. We began by attempting to forget our previous experiences of reading Petronius and strove to construct a new "first reading." While our construct will not correspond in every particular to everyone's or anyone's first reading of the Satyricon, this is a useful exercise nonetheless, for it helps us to remember and revitalize aspects of the reading experience we regularly choose to forget.

We have thus used the axis of time to unify our experience of reading. If nothing else, the first-person voice of the narrator has also raised in our minds as readers an expectation of unity. Yet we have encountered and recorded in detail in our first reading the tendency of the actual experience of reading the Satyricon to fragment—stylistically, generically, and socially as well as physically. That very fragmentation of experience now invites, even more demands, that we become re-readers of the Satyricon—but what kind of re-readings are licit? What approaches can supplement but not violate our initial framework? What detective games can we play without abolishing the temporality of the text?

One strategy of re-reading, which does attempt to respect this temporality, confronts and seeks to abolish the physical fragmentation of the text by playing ever more intensely the game of reconstructing the missing bits. In our first reading we have done some minimal reconstruction work, though an attempt has been made to avoid digressions from the existing text. There is a challenge to reading an only partially surviving work. Experienced readers of classical texts are familiar with this challenge; we are used to lost books of Livy and missing acts of Menander. That very familiarity is, however, a danger. It conditions us to read in certain ways, and especially encourages us to play the game of filling in the gaps in these texts in time-honored fashion, promising honor (what Stanley Fish has so candidly called "the profession's highest rewards")¹ to those who offer the most compelling reconstruction of the missing bits. Clues within the text become tools for filling in gaps in narrative. ("What is Encolpius doing in Marseilles? What is the chronology of the Quartilla episode?") The problem with this approach pursued alone is that we further disorder an already damaged text without necessarily haveing understood its original order. We may thus gather together all the scattered references to Lichas, Tryphaena, and Lichas's wife in order to reconstruct the missing prehistory of Encolpius's relations to these three, but we have by no means thereby "explained" the episode on Lichas's ship as we now have it.

This problem seems particularly acute for the Satyricon, where the evidence suggests that we have at least thirteen books to reconstruct before our fragments even begin. Doubts about what might have been in the lost books might have paralyzed interpretation, had not trained readers of ancient texts a method of reading at their disposal to assist in the reconstruction of the lost parts of the work, even in the absence of explicit references to those missing portions in the surviving text. That method is generic criticism.

This is indeed a legitimate way of re-reading certain classical texts. We know enough about Menandrian comedy and the conditions of its creation, performance, and reception to allow the construction of a generic archetype. Surviving plays show enough structural features in common that we can say with some certainty what the end of the fourth act of a Menander play is likely to be or who of a given cast of characters is likely to be introduced in the first act. The hope of constructing such an archetype for the Satyricon explains many generic approaches to the work.

But is this a legitimate way of re-reading the Satyricon? It is one of only two surviving Roman novels, and the other, the Golden Ass, does not offer us a model for reconstructing Petronius. The problem of parody enormously complicates the generic approach. Which generic model shall we look to? Were there, for example, other explicit Odyssey references, parallel to Encolpius playing Polyaenus with Circe, which would "prove" that the Satyricon was a parody of epic? Were there other banquet scenes and parodies of Roman social customs which would conclusively make the work satire?

I shall not answer these questions; instead, I intend to examine the premises behind them. The assumption seems to be that the genre of a work is determined by the proportion from a certain language system in the full conglomeration. If more than 50 percent of the language is that of epic, then the genre is epic, or at least parody of epic. This assumes that the fundamentally mixed nature of the existing text of the Satyricon can be tamed or disciplined by the recovery of a more unitary body of missing language. I deny this assumption.

Part of the reason for insisting on the linear experience of our first reading has been precisely to recover the sense of literary mixture in the Satyricon. A typical generic approach begins by radically disordering the text along the axis of time. It may, for example, take the references to Priapus from the end of the Satyricon and read backwards, looking only for bits of epic parody in the text which can be said to lead up to the "theme" of the wrath of Priapus. This approach demands that the reader become an archaeologist of, or, perhaps better, a detective within, the text.

The model of the reader as detective is already a familiar one, but nonetheless useful. Many fictions seem to invite the reader to operate as a detective, or more precisely to operate as the fictional archetypes in detective stories do. John Winkler has discussed this analogy in illuminating detail for the later Roman novel of Apufeius,² and we may yet discover that Petronius is playing a kindred game. The unspoken assumption is that a text of literary quality is "difficult"; it possesses a meaning that must be wrested from it by the power of the reader/detective over the text. Some approaches assume that the text is composed according to the "fair-play" rules of detective fiction.³ All of the facts are there. It is up to the reader/detective to discard irrelevancies and red herrings, rearrange the chronology of presentation into a chronology of action (as with alibis based on time), and produce (for the read.er's pleasure) the truth, the hidden reality behind the rhetoric of seeming. Other, more deconstructionist, approaches assume we can insert a lever and pry loose the meaning the text wishes at all costs to conceal.

The reader/detective thus creates a hierarchy within the text as he reads. A generic approach based on the epic model, at its extreme, "reveals" to us a hidden text, made up only of epic and Priapic references, which we presume Petronius has intentionally buried under all his other material. The remaining poems, folktales, situations from mime, or other parodies become filler or, even worse, a collection of red herrings designed as a test, devices to divert us from the supposed buried treasure the author has created for the reader/detective to ferret out, the "real meaning."

No doubt some texts, especially those based on parody of, or allusion to, a single other work of literature, are composed on such principles. Even there, an analysis that simply extracts the allusions does justice to neither the author's rhetoric of presentation nor the reader's reception of the text. In such texts, however, once the hidden armature of the text parodied or alluded to is revealed, its existence is usually indisputable. The fact that so many texts or genres are parodied or alluded to in the Satyricon suggests there is no such hidden armature, or at the very least no hierarchy in which one parody is central and all others peripheral.

We must rather confront the mixture of languages, subjects, and style in the Satyricon as we have it. Nor is it enough to label the mixture prosimetrum or Menippean satire and declare that we have solved the generic question; the labels simply beg the question. What is the effect of this continual shifting among kinds of language as a reading experience?

A recognition of this mixture of language systems, which he terms heteroglossia, is the foundation of M. Bakhtin's approach to the novel. In his view the novel itself is opposed to genre; it has no genre itself but coopts into itself the other literary genres (notably epic, but also satire, lyric, etc.). Bakhtin then subdivides the novel into two "lines." In the first, the novel takes the subjects and concerns of the other genres, but reworks them in a new, spuriously "unified," literary language. As he says: "Novels of the First Line, as we have seen, incorporate a multitude of different semiliterary genres drawn from everyday life, and proceed to eliminate their brute heteroglossia, replacing it everywhere with a single-imaged, 'ennobled' language."⁴ One might say that the novel absorbs a class of signifieds from the various genres while using and creating a new and apparently unified set of signifiers to discuss these matters. For example, the epistolary novel brings into literary prose a range of subjects and narrative patterns new to high literature, but they are absorbed into, and represented by, the exiting literary language. In the Second Line, the newly incorporated genres bring with them their distinctive signifiers as well. In Bakhtin's terms, a genuine sense of dialect becomes apparent; texts speak in two or more distinctive voices, and the linguistic texture of the novel becomes dialogized.

Our two surviving Roman novels fall linguistically into these two lines: it is apparent that there is a characteristically Apuleian style, whereas there is no single Petronian style. The rich and variegated language of the Golden Ass is nonetheless shared by characters of widely different social and generic backgrounds: the maid Fotis speaks a language as richly literate as that of Lucius himself. In the Satyricon, however, language is a good indicator of social status, and a character such as Circe's maid, Chrysis, who speaks in a literate vein despite her status, is quite unusual.⁵ In Petronius the various language systems are not submerged and absorbed into a unified literary style. Encolpius on the decline of rhetoric sounds nothing like Echion (45) on the decline of gladiatorial fights. Put another way, the freedmen's solecisms are grammatical, our heroes' those of taste.

On a crude level this is one explanation for the "modernity" so many readers have experienced in the Satyricon. Bakhtin's distinction between First and Second Line novels is not strictly a chronological one, but First line novels, as represented by romance and chivalric fiction, are generally prior to linguistically and socially heteroglot fictions. The matter of the genres mixes before their media do. The Satyricon is thus a novel very much before its time.

Auerbach's brilliant essay on the Satyricon in Mimesis sees this change from the unified language of epic to the heteroglot language of Petronius as a key advance in realism: "Petronius' literary ambition, like that of the realists of modern times, is to imitate a random, everyday, contemporary milieu with its sociological background, and to have his characters speak their jargon without recourse to any form of stylization. Thus he reached the ultimate limit of the advance of realism in antiquity."⁶ The Satyricon has thus taken the same step beyond the unified literary language of the Greek romance that the later novel (the preeminent example being Don Quixote) takes beyond chivalric and sentimental romance.⁷

The Satyricon in turn makes this unified language of romance one of the objects of its own representation, a development key to Bakhtin's Second Line novel:

As a counterweight to "literariness," novels of the Second Line foreground a critique of literary discourse as such, and primarily novelistic discourse. This autocriticism of discourse [italics Bakhtin's] is one of the primary distinguishing features of the novel as a genre. Discourse is criticized in its relationship to reality; its attempt to faithfully reflect [sic] reality, to manage reality and to transpose it … And in its further development, the novel of the Second Line remains in large measure a novel that tests literary discourse.⁸

This objectification of the prior novel form serves not only the purposes of realism (as we shall see), though that is an important aspect of it: there are literary people in the world who can only be represented in this way. The process also has metafictional implications for other literary forms represented within the Satyricon and for the characters who introduce this "literariness" into their own discourse.

I propose to begin re-reading the Satyricon by examining in turn the various language systems that have gone into its construction. Here I may seem to violate the axis of time so important to our first reading by gathering examples of language from the whole work by social stratum or generic affinity. This violation is more apparent than real. Unlike the approaches I have criticized, in theory all the language of the Satyricon can be reexamined in this way; no preliminary hierarchization takes place such as occurs when we reread with a single generic model in mind. Moreover, a language system in Bakhtin's terms is a synchronic entity. In examining the range of expression from a given language system actually used in the Satyricon, we are therefore simultaneously exploring the repertoire of the reader implicated in this text. The repertoire does not come into being along the time axis of the reading process, but in theory exists in the mind of the reader from the beginning. An example is in order here. At 29.1, Encolpius is frightened by the painted dog into a pratfall. At 72.7, a real watchdog frightens Ascyltus and Encolpius as they are trying to escape from Trimalchio's house, and both end up in the fishpond. These two incidents are part of a ring structure. Note, however, that the second event simply does not exist when the reader encounters the first. Whatever the first encounter with a dog means, its meaning is not "explained" by the second encounter. That is only a meaning added to the meaning of the original encounter in its original context. On the other hand when a word such as versipellis occurs in Niceros's story of the werewolf in sections 61-62, we can presume this concept to be part of the linguistic universe of the reader of the Satyricon from the beginning. The language system of ghost and horror stories from which this comes is therefore potentially present throughout the novel.

How do we establish the occurrence of various language systems in the Satyricon? Individual instances are established by difference. The formal markers of change are often obvious: for instance, from Eumolpus reciting hexameter poetry to Encolpius describing events on the road to Croton, or from Agamemnon summarizing a controversia for Trimalchio's amusement to Niceros's quite different entertainment in the form of the werewolf story. Yet this leaves us with a welter of instances and no general principle as yet for dealing with them.

Only by gathering together instances of a given language system from the whole of the Satyricon are we enabled in turn to judge the impact of a language system in its context and to relate the instances of a given language system to its use in Roman culture at large in Petronius's time. Each instance of a language system has at least three valences, three horizons to read against. The first is its immediate context, the alien language system it is placed next to in the narrative of the Satyricon. Our first reading has given particular attention to this valence. The second is to other, noncontiguous, examples of that language system in Petronius. The third is to the language system as a whole and its place in the discourse of Roman culture outside the given text. As we fragment the text of the Satyricon into its constituent language systems in the course of our re-reading, we shall focus particularly on these last two valences.

Yet considerable problems remain in defining individual language systems. Shall we group system by speaker, by linguistic form, or by function in the narrative? Our first reading has addressed the marked shifts of language within the narrative. For our re-reading to enrich, rather than simply suppress, the experiences of our first reading, we must look at as many potential language systems as possible. Two somewhat different approaches to the heteroglossia of Petronian narrative are implicit in the brief excerpts quoted above from Auerbach and Bakhtin himself. Auerbach sees the heteroglossia of Petronian narrative in social terms: the speech of slaves and freedmen versus that of free citizens, the uneducated versus the educated. Bakhtin sees this already filtered through literary forms. Auerbach describes Petronius's language as "jargon without recourse to any form of stylization"; Bakhtin suggests that a stylization has already occurred in such minor literary forms as mime or epistolography. We must beware simply collapsing these categories. The "semi-literary" of Bakhtin's hierarchy need not be identical with the "unstylized jargon" of the freedmen and slaves. Bakhtin's classification in a sense is nearer to what literary history and our close reading of the Satyricon have already suggested: the speech of the freedmen, for example, is not simply a transcription from nature, but shows considerable influence from such popular forms as mime. Yet Bakhtin himself refers to the passage cited above to "brute heteroglossia" which is "drawn from everyday life." The effect certainly is to conceal, not reveal, any literary (in our sense of the word) origin for certain language systems in Petronius.

Insofar as social and literary classifications of the Satyricon's heteroglossia are in competition, we need not adjudicate their claims beforehand. Both ways of reading will help us hear the polyphonic nature of this remarkable text more clearly. I propose to begin by examining the social and character differences encoded in its language and shall turn in the next chapter to the literary voices of the text.

In beginning our re-reading of the Satyricon with a search for social difference in language, we must beware a certain danger. Petronius's characters, especially the freedmen at Trimalchio's table, are so different from the rest of Latin literature and so individually memorable that it is tempting to proclaim each an island unto himself and assume beforehand that their character differences are expressed in the form as well as the content of their language. My goal at the moment, however, is not to offer a full study of characterization in Petronius; character is the sum of many language systems, not just the social. We must begin more simply, with the way the characters speak.

Social difference in language in the Satyricon paradoxically is at once both obvious and elusive. At first glance it is clear that the Latin of the freedmen at Trimalchio's table is neither that of Vergil and Cicero nor that of the work's more educated characters. Early philological interest quickly focused on the evidence the text provided for the distinction between sermo urbanus and sermo plebeius (roughly, literary and colloquial language). Vocabulary and idiom, the variations from standard genders or inflectional endings, sentence structure and rhythm of speech were all investigated carefully—and, supplemented by comedy, inscriptions, and other scattered sources, yielded a good bit of information.⁹

Yet it is difficult to refine this distinction of sermo plebeius and sermo urbanus any further—and indeed we may in fact wonder how representative the freedmen in the Cena are of a Roman sermo plebeius. A glance at their names suggests the problem: Trimalchio, Habinnas, Niceros, Hermeros. These are not native Latin speakers. On the other hand, neither are Lichas and Tryphaena, yet their Latin seems of a piece with the sermo urbanus of Encolpius and Eumolpus. Indeed, the social class of characters other than the freedmen in the Satyricon is hard to judge. One scholar dubbed Encolpius and Ascyltus the Lumpenintelligentsia, which seems at once wonderfully apt and yet still insufficient to fix them on the social scale.

The first-person form of narration also imposes restrictions on our investigation of socially based language systems. We tend to forget how little we have that is not Encolpius—though certainly his own language, both in direct discourse and in narration, is compounded of many systems. We may begin by looking briefly at the language of the freedmen as a single system; thereafter the analysis will turn to the language of individual characters, where distinctions are more those of style than of vocabulary and syntax.

Many features of the freedmen's speech are peculiarities of single words, where it is often difficult to distinguish the errors of the non-native speaker from the uneducated native. Trimalchio (39.5-6) and Echion, for example, both use the form caelus instead of caelum. Seleucus (42.5) and Trimalchio (71.1, 77.2) also speak of "fate" asfatus rather than fatum. While these might in isolation seem errors of one who has learned Latin as a second language, there is ample evidence (including some changes to feminine forms) for the tendency of the neuter to disappear in colloquial Latin.¹⁰ All the instances of changes of gender in the Satyricon occur in the Cena.¹¹

Other peculiarities seem specific to lower-class language. The freedmen are fond of coining adjectives with grand-sounding endings, such as -ax and -osus: for example, abstinax (42.5), nugax (52.4), salax (43.8), dignitosus (57.10), and sucosus (38.6).¹² A contrary tendency, real or specious modesty, appears in the extensive use of diminutives, a common feature of informal speech at many social levels. The tone often verges on sarcasm. When Hermeros attacks Ascyltus for laughing at himself and his fellow freedmen, there is considerable pride in this statement that

glebulas emi, lamellulas paravi …
(57.6)

I bought a little land, I got together a little
  silver …

and when he challenges Giton to a little bet (sponsiunculam, 58.8), the diminutive suggests his contempt for his opponent. Certainly diminutives are not unique to the speech of the freedmen, but they are more common here than elsewhere.

Pecularities of syntax worth noting include the tendency to use the accusative even when another oblique case would be expected (as at 44.16, meos fruniscar) and the use of quod and quia clauses where one would still expect the infinitive plus accusative construction in indirect speech.¹³ These, too, are tendencies that triumph in later Latin.

Turning from vocabulary and grammar, we also find features of style shared broadly among the freedmen. Their speech is notably telegraphic, even for after-dinner conversation. These men speak in short sentences and to the point. Here is Ganymedes complaining about the decay of religion:

nemo enim caelum caelum putat, nemo jejunium servat, nemo Jovem pili facit, sed omnes opertis oculis bona sua computant. antea stolatae ibant nudis pedibus in clivum, passis capillis, mentibus puris, et Jovem aquam exorabant. itaque statim urceatim plovebat; aut tunc aut numquam: et omnes redibant udi tamquam mures.

(44.17-18)

No one thinks heaven is heaven, no one keeps a fast, no one cares a straw for Jupiter, but they all shut their eyes [to everything else] and count their money. Once upon a time matrons used to climb the hill with bare feet, hair down, minds pure, and pray to Jupiter for rain. And it immediately poured down in buckets—either then or never—and they all came back drowned like mice.

Partly the effect arises from the parataxis: strings of phrases tied together with nam and et or no connectives at all.¹⁴ Partly also it is a result of the liberal use of proverbial expressions and cliche phrases (such as udi tamquam mures and urceatim plovebat).

Such proverbs and cliches are virtually a language system of their own. A phrase such as serva me, servabo te (44.3: "You scratch my back, I'll scratch yours") is at once part of everyone's language and no one's in particular. Formal speech and writing avoid such phrases precisely because they are not original or distinctive; they are the values of the culture thinking out loud. When Echion says colubra restem nonparit (45.9; literally, "A snake doesn't give birth to a rope"), he is expressing not so much a thought in proverbial form as a demand for his audience's affirmation of his previous statements. At times these demands and stock phrases threaten to crowd out any real ideas at all.¹⁵ Here is Phileros on the late Chrysanthus:

puto mehercules illum reliquisse solida centum, et omnia in nummis habuit. de re tamen ego verum dicam, qui linguam caninam comedi: durae buccae fuit, linguosus, discordia, non homo. frater ejus fortis fuit, amicus amico, manu plena, uncta mensa. et inter initia malam parram pilavit, sed recorrexit costas illius prima vindemia: vendidit enim vinum, quantum ipse voluit.

(43.2-4)

By Hercules I think he left a hundred thousand minimum, and all of it in cash. And I'm telling the truth about this: I've eaten the dog's tongue:¹⁶ mean-mouthed, talkative, an argument, not a man. His brother was a real man, a friend to a friend, free with his money and a good host. And when he was starting out he had some bad luck, but his first vintage patched him up: he sold his wine at whatever price he wanted.

Between Phileros's insistence on his own truthfulness and his rhythmic, but stereotypical, paean to Chrysanthus and his brother, there is hardly room for any real narrative or description, which boils down to this: Chrysanthus, a hard man in contrast to his brother, had reverses and successes. One imagines that Phileros describes anyone he approves of as an amicus amico, for example.

Some differences among the individual freedmen make themselves felt, of course, but these are expressed more often through the content and occasion than the form of their language. The first speakers after their host's temporary departure in 41.9 are, like Trimalchio, also haunted by the theme of time. Dama complains that the day goes by so fast that "you turn around, and it's night" (dum versas te, noxfit, 41.10). Seleucus tells of the funeral he has just attended, and Ganymedes (a wonderfully ironic name for the man who sounds the oldest of all of them in his outlook) goes on about the days when men were men and women were devout and chaste. This group of freedmen seems in general older and less directly dependent on Trimalchio himself. Later we meet Niceros, whom Trimalchio can call on for a rendition of the werewolf story (61-62), and Plocamus, who gives an unintelligible performance in Greek. These, like Hermeros, who leaps to Trimalchio's defense when he thinks Ascyltos and Giton are laughing at him (57-58),¹⁷ seem more directly dependent on Trimalchio. There are, then, differences of status within the group of freedmen, but these are not marked by form of speech.

The effect is of a language system independent of the individual speaker, encoding the freedmen's basic values (individualism coupled with strong class loyalty). Consequently it is difficult to distinguish between individual freedmen on the basis of language: their much-praised realism consists more in their deviations from the sermo urbanus than in distinctions among themselves.¹⁸

Our task of listening for individual voices is not simplified as we turn from the freedmen to the other characters of the novel. In a deservedly well-known article of 1966, Peter George undertook an analysis of the style of the non-freedmen as an expression of their character. It is no disparagement of his work, however, to note that he showed more clearly what these characters' style was not: their own. George begins with a discussion of Giton's style and concludes, quite rightly, that it is the style of the declamation exercise, all epigrams and dramatic poses. Moreover it is bad rhetoric, an expression of Giton's immature and undisciplined character.¹⁹ Paradoxically though, Giton's style is defined by its lack of personality: it is his insofar as it is not his. Where the freedmen lose themselves in the cliches of everyday life, Giton loses himself in the conceits of rhetoric. So, too, George's discussion of the maid Chrysis's style merely compounds a negative definition: in one passage she speaks in the language of the slave-and-freedmen social stratum to which she belongs, while in another she apes the rhetorical mannerisms of the social-climbing set of Encolpius, Ascyltus, and Giton.²⁰ Neither style is her own in any real sense. Although the rhetorical style is clearly adopted, the more colloquial style is not marked by any personal idiosyncrasy and so blends with the rest of the sermo plebeius in Petronius.

Eumolpus would seem the most promising candidate for an analysis of his personal voice, in that he speaks more of the Satyricon than anyone other than Encolpius. While the bulk of this is in the form of his two poems, the Bellum civile and the Trojae halosis, which present their own stylistic problems, there is still ample material from prose contexts to consider. Yet a distinctive voice for Eumolpus remains elusive. There is no stylistic overlap between his poetry and prose; in fact, George notes that "his prose is less colored by poeticisms than that of any of the other literati."²¹ George admires Eumolpus's style and discusses how his character is the appropriate narrator for the Milesian tales in the Satyricon, but then he goes on to say that Eumolpus's style is "notable for just that blend of literary vigor" that characterizes the best prose in Petronius.²²-That means Encolpius's style.²³ In other words, Eumolpus's style is just the same as Encolpius's style, absent the vices sporadically "characteristic" of the latter. Thus Eumolpus emerges just as stylistically schizophrenic as Chrysis, though where she divides between the sermo plebieus and an attempt at sermo urbanus, Eumolpus divides between a polished sermo urbanus, which he shares with the other educated characters (even though he may excel in it) and a poetic style which we have ample reason to believe is not his own (and shall return to below).

When confronted with the challenge of identifying a characteristic voice and language for Encolpius, the immediate response is to throw up our hands. It is not merely that, since he is the first-person narrator, technically every word of the Satyricon is in "his" voice. Our situation is not materially improved if we limit our consideration to the language Encolpius speaks out loud in the course of the novel. As George notes:

Encolpius's style of speech shows no slavish adherence to one particular model or set of models: if it did, the sheer quantity of what he says would make for monotony. But this does not mean that Encolpius is less imitative than Giton; on the contrary. Many of the variations in his style are ad hoc imitations of the style of the person to whom he is speaking at the time.²⁴

Thus Encolpius, somewhat more successfully than Giton, speaks the languages of others; it is an intensely literary style.

This brief survey has suggested that there is less than meets the eye to the social and character differences of language systems in this novel. The broad distinction of sermo plebeius and sermo urbanus suffices to generate the heteroglossia of sociological background that Auerbach praises, but it is much more difficult to trace finer social distinctions of speech. Variations within the language of the more educated characters are more a function of the literary models they use or misuse than characteristic expressions of individual psyches. Our next step must therefore be an exploration of the literary language system of the Satyricon.

Notes

¹ Fish 1981, 371.

² Winkler 1985, esp. 60-69. My discussion of reading and detective fiction owes a great deal to his. See also Guetti 1982.

³ S. S. Van Dine's list of twenty rules is the classic, first published in the American Magazine, September 1928 (conveniently reprinted in Haycraft 1946, 189-96, which also contains Ronald Knox's "Decalogue," 194-96).

⁴ Bakhtin 1981, 410.

⁵ Many words appear nowhere in Latin literature between Plautus in the second century B.C. and Apuleius in the second century A.D. In his excellent analysis of Apuleius's style, Tatum 1979, 147, notes the Plautine exuberance in Fotis's style. He further notes: "Although stylistic variation abounds in Apuleius' work, his novel never attains the variety of characterizations so distinctive in Petronius' Satyricon. The qualities of Petronius' characters are reflected in their speech … Such verisimilitude never appears in Apuleius" (149). On Chrysis's style, see George 1966, 342-46.

⁶ Auerbach 1953, 30. He does regard the Satyricon as not fully realistic in the modern sense because of its failure to anchor this mixture of languages in a concept of historical forces that produced them (31-33).

⁷ It is, in point of fact, earlier than the surviving examples of the Greek romance. New discoveries could easily revise our picture of the history of ancient prose fiction. At the moment, though, the essentially unified language of Longus or Heliodorus seems typical of the Greek romance.

⁸ Bakhtin 1981, 412.

⁹ The discussion of the sermo urbanus and the sermo vulgaris or sermo plebeius begins with Buechler in his edition of 1867. One of the most important works here is Abbot 1907. Smith 1975, xxix-xxx, provides an excellent short bibliography on linguistic questions and a very useful outline of the linguistic idiosyncracies of the freedmen in his appendix 2 (220-24). What follows is deeply indebted to Abbott and Smith. The title of Petersmann 1985 promises a great deal more in this area than it delivers. Petersman 1986 offers much the same material, but the sections on sermo urbanus and sermo vulgaris (401-7) are useful. A copy of Heraeus 1899 came to hand too late to be incorporated in this analysis but contains important parallel materials.

¹⁰ Smith 1975, 221.

¹¹ Swanson 1963, 253.

¹² Smith 1975, 222-23. Abbot 1907, 46, also notes the use of the popular endings -arius and -atus.

¹³ E.g., dixi quia mustella comedit, 46.4.

¹⁴ Abbott 1907, 48.

¹⁵ Abbot 1907, 48, counts seventy-five such examples in the space of four pages of text.

¹⁶ Meaning obscure. Suggestions range from a medicinal plant (Heseltine) to a reference to Cynic philosophers (Sage and Gilleland, ad loc.).

¹⁷ Hermeros in this passage is given one touch that is emphatically not freedman's speech. He imagines Ascyltos asking him a question with the word servivisti in it: "Why are you a slave?" The normal form of the verb was contracted (servisti). This is the only uncontracted -vis- perfect in Petronius. Vine 1988, who points this out, suggests Hermeros is mocking the hypercorrect speech of intellectuals and characterizing Ascyltos as such. That is too subtle a touch for Hermeros's character; Petronius happily sacrifices consistency of character here for a good joke.

¹⁸ Abbott 1907, 49, argues for some individuality: seven of eleven "plebeian" words in his sample are spoken by Echion; Phileros is fond of the oath mehercules. Smith 1975, 223, notes the prevalence of Graecisms in Hermeros's speech.

¹⁹ George 1966, 338-42.

²⁰ Ibid., 342-46. The colloquial passage cited is 126.8-10, where she rejects Encolpius's advances; the more rhetorical is 126.2.

²¹ George 1966, 348. Cf. Gagliardi 1981, 364: "Tra Eumolpo poeta ed Eumolpo personaggio, in somma, si ricompone qui una frattura altrove sempre operante."

²² George 1966, 347.

²³ George indeed believes that one can distinguish Petronius's style from Encolpius's style, despite the first-person form of narration. This procedure seems to me merely to divide the prose one admires ("Petronius's style") from everything else ("Encolpius's style") on no principal other than taste. Roger Beck 1975 has attempted a functionally similar, but unpersuasive, division of Encolpius into younger and older selves. The recent attempt of Jones 1987 to refine Beck's approach in fact demonstrates its unworkability. See above ch. III n. 26 and ch. VI n. 33.

²⁴ George 1966, 350.

Works Cited

Abbot, Frank Frost. 1907. "The Use of Language as a Means of Characterization in Petronius." Classical Philology 2: 43-50.…

Auerbach, Erich. 1953. Mimesis. Translated by W.R. Trask. Princeton.…

Bakhtin, M. M. 1981. The Dialogic Imagination: Four Essays. Edited by M. Holquist. Translated by C. Emerson and M. Holquist. Austin.…

Beck, Roger. 1975. "Encolpius at the Cena." Phoenix 29: 271-83.…

Fish, Stanley. 1972. Self-Consuming Artifacts. Berkeley and Los Angeles.

——. 1980. Is There a Text in This Class? The Authority of Interpretive Communities. Cambridge, Mass.

——. 1983. "Profession Despise Thyself: Fear and Self-Loathing in Literary Studies." Critical Inquiry 10: 349-64.

——. 1984. "Fear of Fish: A Reply to Walter Davis." Critical Inquiry 10: 695-705.

n. 1985a. "Anti-Professionalism." New Literary History 17: 89-108.

——. 1985b. "Resistance and Independence: A Reply to Gerald Graff." New Literary History 17: 119-27.

——. 1987. "Change." South Atlantic Quarterly 86.4: 423-44.…

George, Peter. 1966. "Style and Character in the Satyricon." Arion 5: 336-58.…

Guetti, James. 1982. "Detective Fiction." Raritan 2.1: 133-54.

Haycraft, H. 1946. The Art of the Mystery Story. Reprinted New York, 1975.…

Heraeus, Wilhelm. 1899. Die Sprache des Petronius und die Glossen. Leipzig.…

Heseltine, M. 1969. Petronius. Revised by E. H. Warmington. Cambridge, Mass.…

Jones, F. 1987. "The Narrator and the Narrative of the Satyricon." Latomus 46: 810-19.…

Petersmann, H. 1985. "Umwelt, Sprachsituation und Stilschichten in Petrons 'Satyrica."' ANRW II. 32.3: 1687-1705.

——. 1986. "Petrons 'Satyrica."' In Die romische Satire, edited by J. Adamietz, 383ff. Darmstadt.…

Sage, E. T. 1969. Petronius: The Satyricon. Revised and expanded by B. Gilleland. New York…

Smith, Martin S., ed. 1975. Petronii Arbitri Cena Trimalchionis. Oxford.…

Tatum, James. 1979. Apuleius and the Golden Ass. Ithaca, N.Y.…

Vine, Brent. 1988. "A Note on servivisti (Petr. 57.4)." AJP 109: 543-46.…