Beyond the Second Sophistic. Tim Whitmarsh
a classicizing world, sometimes an explicitly classical one (in Chariton and Heliodorus). As has often been noted, moreover, the romances recycle a number of set-piece topoi: love at first sight (preferably at a festival), separation, kidnap by pirates, intense experience of conflicting emotions, the false appearance of death (Scheintod), courtroom scenes.3 Further evidence for genericity can be sought in the titling conventions, which (I have argued) take a distinctive form: “Events concerning [ta kata or ta peri] x girl,” or more usually “… x girl and y boy.”4
Because of the relative consistency of the form over some three hundred years, critics have sometimes presented the generic identity of the romance in terms of adherence to a schematic narrative template.5 This approach, however, risks downplaying the degree of variation. Each of the surviving five romances is actually very different: if Chariton, quite possibly the earliest, represents the “norm,” then Xenophon contrasts with his low-grade style, Achilles with his first-person narrative and emphasis on gore and lechery, Longus with the pastoral setting, and Heliodorus with his sanctity and African location. With repetition of narrative motifs, moreover, come improvisation and variation too: so, for example, Longus’s miniaturized pastoral romance has a failed kidnapping in which the abductors do not make it out of the bay (1.30.2), while Achilles’s exuberantly over-the-top text features three different false deaths (each of which his credulous protagonist and narrator believe in), and so forth. Genres are not schematic; more recent commentators have, instead, preferred the language of “family resemblance,” a Wittgensteinian term first applied to genre theory by Alastair Fowler in 1982.6 Members of a family are often visibly identifiable as related without sharing identical features; the same model might be used for the romances. The family analogy is also useful in that it gives a role to genetic admixture. Families, if they are not entirely incestuous, propagate themselves by mixing in new DNA; similarly, new texts within literary genres show difference as well as sameness.
This model, however, raises new problems. Families are social constructs rather than straightforward mirrors of biological truth: not all children are the natural offspring of those whom society recognizes as their fathers and mothers. Similarly, identifying the “ancestry” of literary texts can be a more complex issue than it initially appears. This is all the more so in relation to the imperial romances, which are radically intertextual, cannibalizing other forms voraciously: they absorb features from classical epic, tragedy, historiography, New Comedy, rhetoric, lyric, and so forth. What is more, they have numerous points of contact with other “nonclassical” varieties of contemporary literature: a case in point are Christian martyrologies, which often follow a similar pattern of quasi-erotic infatuation leading to obstacles and challenges and finally redemption (although in the self-denying world of early Christianity, it is death rather than sex that marks the telos).7 For some, the romance’s innumerable points of literary reference point to an absence of coherent generic identity. For Steve Nimis, for example, the romance is “anti-generic, unable to be specified as a single style of discourse.”8 Helen Morales has recently developed this claim at greater length. “The evidence that we have suggests that there was no ‘traditional genre’ of the ancient novel,” she argues, using Anders Petterson’s phrase denoting “a type of literary work which is generally recognised within a culture, as a special type of work.”9 Rather than defining genre in formalist terms, she argues, we should be viewing the “novel” as an “imaginative mode” with various recurrent features: a concern with boundaries and limits, an attempt to map out morality, an opposition between chastity and prostitution. A “ ‘novelistic’ mode of imagination is one that both heightens and exaggerates things, that simultaneously reveres and degrades women, and that suggests that the domestic . . . as opposed to the mythic is a place for the instauration of significance.”10 Once we view “the novelistic” in this way, then we can begin to see new points of connection with, for example, Nonnus’s Dionysiaca and Musaeus’s Hero and Leander, and with Roman declamations.
This approach, I think, offers both opportunities and risks. Opportunities, because it challenges the misleading view that genres are somehow ontologically nonnegotiable, a view that the disciplinary practices of the modern academy perpetuate. How many undergraduate courses on the ancient novel include Seneca the Elder, Christian martyrology, and Nonnus? Yet as Morales rightly observes, there are all sorts of points of contact between these different works, and a rich cultural history of the imperial age would need to map out the contraflowing traffic between these many different types of text. That very formulation, however, points to the problem: we need to account for difference as well as identity, for it is intuitively implausible to imagine that ancient readers would turn from Longus to a declamation to a hexameter epic without registering any generic jolt. It might be countered that this jolt occurs because of the generically determined nature of declamation and epic and not of novel/romance, which lacks decisive formal (e.g., epic’s meter and diction) and contextual markers (e.g., a sophistic auditorium, in the case of declamation). Here I would agree up to a point: novelistic writing in general is uniquely fluid and multifarious, generically speaking. But within the broad category of “the novelistic,” the romance is, I think, coherently generic. This, indeed, is precisely why we can feel the hybridity of an erotic epic like Musaeus’s Hero and Leander, which imports motifs (and the titling convention) from the romance:11 the effect depends on the reader’s ability to perceive that process of generic cross-pollination, which itself implies an awareness of romance as a distinct literary identity. My argument, then—yet to be fully substantiated—is that Morales is right about “the novelistic” as a general category, but that romance operates according to tighter generic rules. This does not mean that there are no ambiguous cases: clearly Iamblichus’s Babyloniaca, for example, was a heterosexual romance but also a radical experiment in setting and content (and length). But as with Hero and Leander, the fact of the Babyloniaca’s generic experimentation reinforces the argument that there was a genre to experiment with.
My discussion above has, in fact, effected a small but significant shift in definition. What Morales resists, quite rightly, is a conservative, rigid, formalist conception of genre. Yet genre should not be thought of in this way, as an intrinsic property of individual texts, like a gene that can be sequenced; it is, rather, a relationship between texts, a relationship invoked for specific, tactical reasons and to shape the reader’s literary reception of the work in question. It is—this is Fowler’s central point—a communicative device rather than a classificatory one.12 Indeed, it might be said that genre is essential to all human communication, to the extent that (as Mikhail Bakhtin argued) speech has its own genres, each with their own sets of expectations that can be met, intermixed, flouted, or rewritten (greeting, thanking, joking, etc.)13 Literature, in a similar but arguably much more complex way, rests on a contract of accepted rules between author and reader: a contract that is unwritten, certainly, and can be reneged on or rewritten, but is always there. It is this contract that dictates whether a particular action or utterance within a text is received as vraisemblable or transgressive. “Our intuitive sense of this vraisemblance is extremely powerful,” writes Jonathan Culler. “We know, for example, that it would be totally inappropriate for one of Corneille’s heroes to say, ‘I’m fed up with all these problems and shall go and become a silversmith in a provincial town.’ Actions are plausible or implausible with respect to the norms of a group of works.”14 Corneille’s fed-up hero would be acting in much the same way as a real-life person who, when offered a hand to shake, responded with a punch to the belly: both would be in effect breaching a generic contract (or perhaps in the second case refusing to accept one).
Let us at this stage dispose of a potential objection. It is true that extant Greek lacks any consistently attested word for designating the ideal romance.15 In fact, there is not even any consistent word for novel: the best candidate, dramatikon (diēgēma) (dramatic [story]), does not appear before Photius in the ninth century C.E. and even then seems to refer to the “dramatic” aspects of the plot (sufferings and reversals of fortune) rather than to anything distinctive to this kind of text.16 At first blush the absence of any name for or theorization of the novels or the romances would seem to support the view that the boundaries of the genre were not clearly defined. Yet we can plausibly explain the absence of any explicit label in our pre-Byzantine sources by other means. As one recent critic has observed