CAN FIGURES PERSUADE? ZEUGMA AS A FIGURE OF PERSUASION IN LATIN

Abstract Use of rhetorical figures has been an element of persuasive speech at least since Gorgias of Leontini, for whom such deliberate deviations from ordinary literal language were a defining feature of what he called the ‘psychagogic art’. But must we consider figures of speech limited to an ornamental and merely stylistic function, as some ancient and still many modern theorists suggest? Not according to contemporary cognitive rhetoric, which proposes that figures of speech can play a fundamentally argumentative role in speech by evoking a level of shared meaning between speaker and listener, and simultaneously by affording the possibility of reorganizing this common ground. This paper argues that, in Latin literature, zeugma—the ‘linking together’ of two elements (usually nouns or prepositional phrases) with a third (usually a verb) that is semantically compatible with only one of them—can and very often does operate argumentatively, and that it does so by surfacing figurative relationships that normally remain below the conscious awareness of Latin speakers and by imparting a certain structure to these relationships. What very often motivates the selection of elements within zeugma—and what makes zeugma more than simply a stylistic device—are in fact metaphorical structures that are highly conventionalized in Latin's semantic system. In tapping into symbolic associations that are deeply entrenched in the language and thought of Latin speakers, zeugma therefore provided a ready-made device for constructing arguments in context.

cognitive rhetoric, figures of speech play a fundamentally argumentative role in discourse by evoking entrenched conceptual relationships, as well as by imposing a particular structure on these relationships. 4 In this paper, I offer zeugma as a case study of how figures of speech can operate along these lines. Through an analysis of three zeugmatic 'themes' in Latin literature, I show that what licenses the selection of semantic elements within zeugma-and what makes zeugma more than simply a stylistic device-is metaphorical associations that are highly conventionalized in Latin and that therefore provided authors with ready-made and natural-seeming symbolic associations for use in constructing arguments in context. I. PROBLEMATIZING ZEUGMA Ancient grammarians define zeugma as a figure of syntax in which one grammatical element, usually a verb, governs two or more additional elements-usually nouns or prepositional phrases-for each of which the verb is to be understood in a somewhat different sense. Thus Donatus writes that zeugma est unius uerbi conclusio diuersis clausulis apte coniuncta 'Zeugma is the finishing of a single verb by different clauses, fittingly joined to each' (Ars 3.4.5). Similarly, Isidore of Seville states that zeugma est clausula dum plures sensus uno uerbo clauduntur 'Zeugma is a phrase where multiple senses are encompassed by a single verb' (Orig. 1. 36.3). Although the exact nature of the relationship of the governing element to the elements it governs varies from definition to definition (with zeugma being sometimes equated with, sometimes differentiated from, syllepsis on these grounds), generally speaking, this relationship is said to be in one instance semantically fitting and appropriate and in the other (or others) semantically incongruous and irregular. 5 Thus, in modern handbooks, it is possible to find definitions like Ernest Lussky's: 'Zeugma is a figure of speech in which two nouns are joined to a verb strictly suitable to only one of them, but easily suggesting another verb suitable to the other noun'; 6 or, to note one of the most recent, Richard Smith's: 'Zeugma is the governance of two substantives by one verb, or the modification of two substantives by one adjective. The suitability of the verb to each substantive, or of the adjective to each, is in one instance literal, in the other either metaphorical or analogical.' 7 As these definitions suggest, whatever the grammatical category to which the elements of zeugma belong, to be made sense of, the presumed semantically anomalous pairing requires some other than literal process of interpretation. In some cases, the meaning of zeugma may be resolved through analogical suppletion-that is, by providing a semantically more apposite verb for the mismatched term. For instance, in a situation like Sallust's in Iugurtha tantus dolus, tantaque peritia locorum et militiae erat uti … pacem an bellum gerens, perniciosior esset in incerto haberetur 'Jugurtha had so much craft and so much experience of the country and of warfare that it was considered uncertain whether he was more dangerous when waging war or waging peace' (Iug. 46.8), gerens fits idiomatically with bellum, but to resolve its incompatibility with pacem the reader must assume that a different but semantically related verb such as facere has been elided. Alternatively-and also more commonly-the semantic clash may be resolved through figurative interpretation, in which case the governing verb must be read first in a literal and then in a metaphorical sense. So, for example, in Ovid's animaque rotisque | expulit 'He knocked him out of his senses and out of the chariot' (Met. 2.312-13), to correctly interpret the zeugma the sense relation of the governing verb to the two nouns must be understood in one case metaphorically and in the other literally, since, strictly speaking, one cannot 'thrust out' a person from consciousness (anima)-as is literally possible from a chariot, by contrast. Or the relation of the verb to its objects may represent two points on a spectrum of figurativeness, where one relation is nevertheless 'more literal' than the other. For instance, in Ovid's actaque magni | Herculis inplerant terras odiumque nouercae 'The exploits of great Hercules filled the world as well as his stepmother's hate' (Met. 9.134-5), the zeugma hinges on the distance between metonymical and metaphorical readings of inplerant. With terras, this verb has an extended metonymical-but still basically concrete and physical-sense (that is, 'extend over the whole of'), whilst to be interpreted in relation to odium it must be read in a highly abstract and metaphorical sense, something like 'feeding' or 'nourishing' or 'fuelling', since hate cannot literally be 'filled'. 8 As Harm Pinkster has argued, in instances of this latter type zeugma must be seen to 'widen the use of a verb and "create" figurative use'. 9 Attending carefully to patterns of metaphorical expression in Latin reveals, however, that the sorts of symbolic structures that underwrite zeugma in literature are instead narrowly conventional and uncreative. What I mean is that, although zeugma tends to appear in highly stylized texts and to be characteristic of imaginative types of discourse (especially poetry and oratory), understanding the meaning of zeugmatic pairings almost always depends on metaphorical associations that are entirely commonplace and highly pervasive in the language and thought of Latin speakers, rather than on innovative usages whose figurative meaning must be computed within a given local context. 10 This is easily seen in the Ovidian examples just cited: the metaphorical reading of inplerant with a noun denoting an emotional state (in this case odium) is easily inferred, as 'filling' is in fact a conventionalized metaphor for talking about all kinds of emotional experience in Latin. 11 Likewise, the pairing of expulit with anima appears ordinary in a symbolic context where the 'removal' of (more regularly) animus is the normal way of conveying the notion of loss of consciousness. 12 8  In each of these cases, two nouns-one denoting 'courage' (animi or mens), the other denoting 'the hands' (manus, dextra) or, by metonymy for what is held in the hands, 'weapons' (tela)-are paired grammatically with a verb that, strictly speaking, fits only the latter semantically, since there is no literal sense in which an abstract psychological and emotional concept such as COURAGE can 'fall' (cadere). And yet for a Latin speaker the meaning of this supposedly mismatched pairing-namely that the Pierides and the Sabines and Vulcan became disheartened or lost courage-not only would have been easily inferable but also very likely would not have stood out in its semantic aspect as particularly creative, imaginative, or poetic, since evidence indicates that it is instead quite conventional for Latin speakers to talk about courage and its lack in metaphorical terms of 'up(ward)ness' and 'down(ward)ness'.
The metaphorical construal of courage and cowardice in such terms of vertical orientation in fact constitutes a widespread, and systematic, patterning of figurative meaning in Latin. Thus Latin speakers regularly employ verbs referring literally to 'lifting up' (tollere) or 'setting up(right)' (erigere) with animum or animos to convey the meaning we might express in English as 'taking' or 'gathering' or 'plucking up' one's courage or, equally, 'giving' or 'instilling' or 'inspiring' courage. For example, Livy describes Marcellus rallying his men at Nola as literally 'setting up(right) the spirits of his own men' (23.45.5 suorum militum animos erigeret), while Livy's epitomizer conveys the notion that the Parthians took courage from Crassus' death by saying that they 'stood up' their spirits (Ann. Flor. Epit. 4.9 animos erexerant). Correspondingly, words referring literally to 'falling' (cadere) or 'sending down' (demittere) are used in conjunction with animus to convey what we might express again in English as 'losing' or 'giving up' courage. For instance, to express that the Crustumini lost the conviction to go through with the war they had begun against the Romans, again Livy writes that their spirits had 'fallen' (1.11.3 ceciderant animi). Likewise, Caesar uses the construction animo deficere (literally, 'to break down from courage') to express the idea of losing courage in the face of some difficult circumstance, as in 'Marcius Rufus … entreated his men not to break down from courage' (BCiu. An apparent, later imitation of Ovid's cecidere illis animique manusque (in (1)) appears to corroborate the claim that the kinds of metaphorical associations underpinning zeugmatic configurations in Latin literature tend to be ones that are deeply entrenched in the language's semantic system. In rediere uiris animique manusque 'The men [sc. the Argonauts] regained their hands and courage' (Val. Fl. Argon. 4.673), the lexical components of the Ovidian model are partially reconfigured: 14 most importantly, the governing verb cecidere has been replaced by rediere (literally, 'came back, returned'), so that the zeugma is again predicated upon the (presumed) semantic irregularity of employing a verb with an unmistakably spatial sense with animus. 15 However, understanding the meaning of this expression hardly requires an imaginative or creative process of figurative construal, as Latin speakers regularly conceptualize courage in terms of the spatial proximity of animus. For example, when Cicero writes that his arrival in Cilicia greatly emboldened Cassius, he says, literally, that 'At the rumour of our arrival, animus "came near to" Cassius' (Att. 5.20.3 rumore aduentus nostri … Cassio … animus accessit). Similarly, Livy recounts that, when the allied Faliscans, Veientes and Fidenates hesitated to attack the marshalled troops of Mamercus Aemilius before the Battle of Fidenae (426 B.C.E.), 'The dictator and the Romans were emboldened because the enemy had declined battle' (4.18.3 dictatori ac Romanis, quod detrectasset pugnam hostis, animi accessere), and that, while the Romans were concluding a treaty with the Syracusans, Appius stationed a fleet in the harbour 'so as to give courage to those of his own side' (24.27.8 quo suae partis hominibus animus accederet)-in both cases, the act of 'emboldening' being construed literally as animus 'approaching'. In this light, in reworking Ovid's zeugma, Valerius nevertheless relies on a metaphor that is regular within Latin's semantic structures.
Next, consider two very similar passages from Cicero's oratorical works: (4) Q. Metellus … qui de ciuitate decedere quam de sententia maluit (Cic. Balb. 11) Quintus Metellus … who preferred to depart from his city than from his intention; (5) Q. Metellus … cumque in eam legem quam non iure rogatam iudicarat iurare unus noluisset, de ciuitate maluit quam de sententia demoueri (Cic. Sest. 101) Quintus Metellus … when he alone refused to swear by that law which he believed to have been brought illegally, preferred to be removed from the city than from his intention.
The similarity of these passages emerges from certain obvious stylistic correspondences.
To begin with, they exhibit a demonstrable consistency of syntactic arrangement. In each case, the nominative Q. Metellus is followed by two clauses of matching structure. The relative clause with embedded comparative expression (dulcior quam …) in (4) balances with the circumstantial clause with embedded relative clause (legem quam …) in (5). Moreover, unus noluisset echoes conspectus fuit phonically, in its patterning of nasals, sibilants and final stop (ns -(n)st … nsst) and with the repetition of -ui-at the end of the first colon. Most notable, however (and of most interest in our perspective), is the close parallelism of the zeugmatic constructions. Though the lexical components and the word order of these constructions differ somewhat-active decedere varies with passive demoueri, and the position of maluit is reversed in respect to the infinitive that joins the zeugma's constituent parts-in both cases a verb signifying movement down or away from a location (decedere 'depart' or demoueri 'be moved from') governs two prepositional phrases, the first of which denotes a physical space (namely 'the city') and the second of which denotes a mental judgement ('a decision, opinion').
While the similarities in phraseology between (4) and (5) can probably be explained thematically-in both passages Cicero offers a historical exemplum which features Quintus Metellus Numidicus, the famously intransigent opponent of Saturninus' agrarian legislation, and in which the zeugma serves almost as a slogan of aristocratic principled self-sacrifice-comparison with several additional passages again suggests that the zeugmas actually fit an entrenched linguistic formulism. For instance, in (6) and (7)  we find verbs signifying position in a fixed location ( permanere 'remain' and haerere 'hold fast, stick, cling to') joined first with a prepositional phrase referring to a physical space (in urbe 'in the city' or in sedibus 'in their places') and second with a prepositional phrase referring to a mental state of intention (in eadem mente 'in the same intention' or (in eodem) incepto 'in the same purpose'). Although in these instances the notion of MOVEMENT FROM has been replaced by that of POSITION IN, the general semantic configuration of (6) and (7) thus remains largely identical to that of (4) and (5). The identification of all these expressions as instances of zeugma then rests on the (supposedly) anomalous joining of a verb with an unambiguously concrete spatial signification with phrases that denote abstract mental concepts-since obviously one cannot in any real sense 'depart from', 'be moved from', 'remain in', or 'hold fast to' something that exists only in the mind. Yet the metaphorical construal of mental phenomena in spatial terms is hardly imaginative in Latin. Quite the opposite, in fact: this metaphor appears to be a wholly ordinary way for Latin speakers to conceptualize constructs of the mind. Metaphors of spatial motion in fact systematically deliver Latin's ways of expressing concepts of conscious mental attention, according to mappings in which physical movement relative to a location corresponds, through figurative interpretation, to a particular kind of mental operation over some thought. 16 Thus mental operations that involve 'acquiring' a thought in mind-in other words, that involve thoughts either entirely new to the thinker or to which the thinker newly turns his or her conscious awareness-are expressed metaphorically in Latin in terms of movement toward a location in space. So Latin speakers regularly talk about formulating plans, agreeing with opinions, considering ideas, conceiving notions, and so on, as 'entering' (inire), 'occupying' (occupare), 'coming to' (uenire ad …), 'returning to' (redire ad …), 'moving to' (mouere ad …), or even 'falling' (cadere) or 'slipping into' (labi in …) a location. 17 At the same time, words denoting position in a location ('being' or 'standing in') regularly deliver concepts that involve 'having' an idea in mind-that is, mental activity occurring over thoughts as part of the belief system or within conscious awareness. 18 Finally, expressions denoting movement away from a location ('standing away', 'departing from') regularly convey the concept of 'relinquishing' an idea from the mind-that is, giving up some idea that is under current consideration or abandoning some closely held belief. 19 Given the apparently systematic and pervasive nature of this metaphor system in Latin's semantic structures-as well as, indeed, within their symbolic activities at large 20 -instances of this zeugmatic pattern, along with their meanings, therefore appear not only strongly motivated but also to some extent predictable within the context of Latin's overall signifying order.
In these examples, a verb signifying 'extending (forward) (in a determinate direction)'-namely ( pro)tendere-is paired first with oculos, denoting 'the eyes', and then with a noun denoting in one instance a weapon (telum, here contextually understood as 'a bow') and in the other the 'right hand' (dextram). The zeugma is constituted, in one instance, by the literal impossibility of 'extending' the eyes as one 'reaches' the hand, and, in the other, by the distance in figurativeness between 'directing' one's gaze at an object, where interpretation of tendere is highly abstract and metaphorical, and 'aiming' a bow, where this verb has an extended metonymical sense. 21 Once again, however, this (supposedly) semantically mismatched pairing turns out to be motivated by a metaphor that is wholly conventional in Latin, where visual perception may be understood in terms of 'reaching' or 'grasping', as if the eyes were hands. Certainly, in the Latin poets erotic gazing is frequently likened to physical grasping and touching-as, for instance, when Propertius (1. me with her eyes (me cepit ocellis) … who had never before been touched (contactum) by desire'. 22 But the metaphor is more than a poetic image. Indeed, capere oculis (literally, 'to grab hold of with the eyes') is employed by Latin authors as an ordinary synonym for uidere-as in Cicero's ipse accepi oculis animoque sensum 'The sentiment I have seen and understood' (Verr. 2.3. 45) or in Virgil's ante locum capies oculis 'First you will see the place' (G. 2.230).
It is well known, moreover, that a widespread ancient theory imagined vision as a 'ray' or 'beam' of light emitted by the eyes. The Stoics in fact spoke of visual perception as a 'stretching out' of the eye (or its gaze) towards an object (Greek τόνος, ἔντασις; Latin intentio). 23 Apart from its elaboration in the context of expert philosophical theories, this image also appears to characterize ordinary Latin expression. For instance, oculos intendere (literally, 'stretch out the eyes toward') regularly means to 'look at', as when Tacitus (8)) and the hands (as in his oculos dextramque precantem | protendens (9)), the metaphor forming the basis for this expression appears instead to be an entirely regular part of the semantic repertoire of Latin speakers.
For all the instances of zeugma seen so far, then, interpretation appears to depend on metaphorical patterns that are all-pervasive and fully conventionalized in Latin's semantic system. That is, what makes these expressions comprehensible and meaningful is in each case a metaphor that forms part of Latin speakers' automatic everyday speech and thought, and as such constitutes a sort of conceptual background or frame against which the meaning of zeugmatic pairings becomes transparent. But if this is the case, most traditional accounts of zeugma that seek to characterize its effects primarily in stylistic or affective terms, and especially in terms of novelty of expression, would seem to be, at best, too restrictive. Indeed, given the entirely idiomatic, predictable and normal metaphorical modes of expression that seem typically to motivate zeugmatic combinations, it can hardly be that zeugma corresponds to the 'incorrect, audacious, or bizarre combination of words' of, for instance, Dirk Panhuis's recent definition. 24 Likewise, any claim like Heinrich Lausberg's that the semantic inconsistency of a zeugma's lexical components results in a sort of 'tension' that serves to '"create interest" in the linguistic utterance' and 'has the effect of surprise' can only be part of the story, 25 since the perfectly ordinary nature of the metaphors underlying these expressions implies that for Latin speakers no semantic inconsistency existed in the first place 22 The image would later be crystallized by Goethe with his 'see with feeling eyes, feel with seeing hand' (Roman Elegies 5.10). 23 The so-called 'emission theory': T. Rakoczy, Böser Blick: Macht des Auges und Neid der Götter (Tübingen, 1996), 33-5. 24  and so would not have struck them as especially-or indeed at all-interesting or surprising.

II. ZEUGMA AS A DEVICE OF ARGUMENTATION
But if zeugma's function is not-or not only-to endow a certain piece of discourse with some affective quality such as surprise, drawing attention by its unusual semantic configuration, what is its function? Along the lines of contemporary cognitive rhetoric, I would argue that zeugma can and very often does instead operate in a fully argumentative sense, and that it does so by surfacing figurative relationships that normally remain below the conscious awareness of Latin speakers, as well as by imparting a certain character to these relationships. As we have seen, the metaphors that typically underpin zeugma tend to be of an all-pervasive and highly conventional kind, and thus formulaic linguistic expressions reflecting these metaphors likely were not perceived by Latin speakers as figurative at all. Indeed, because in some cases they constitute the inescapable ways of talking about a domain without which Latin speakers would have few or perhaps no other means of doing so, they may even appear to be literal. 26 Zeugma's rhetorical power can thus not rest in its semantic dimension at all. My claim is that zeugma, instead, in purposefully juxtaposing literal and metaphorical interpretations in a single expression, serves to make the normally implicit (because largely automatic and unconscious) figurative association of the relevant domains explicit. In other words, the persuasive value of zeugma has to do with the way in which it brings to overt conscious awareness a particular structure of relations between concepts that normally remains covert. Very specifically, whereas in the metaphorical associations that typically license zeugma inferential patterns belonging to the literal domain are carried over to the figurative domain, in zeugmatic pairings a sort of entailment structure appears to be established between the domains in question, so that the one logically implies the other. In this sense, zeugma can be said to naturalize connections between ideas expressed in discourse. In highlighting the fully conventional nature of the relationship between literal and figurative, zeugma, that is, can make this relationship appear ordinary and expected. And such expectations, I suggest, can be capitalized upon in literature at different levels of argument construction.
Consider again (1), cecidere illis animique manusque. It seems clear that by means of this zeugma Ovid seeks to imply just such a natural link between the Peliades' hands (literally) falling and their animus (figuratively) falling. In this case, the connection Ovid wishes to imply between concepts is a causal one: the girls' hands fall to their sides because they cannot muster the courage to deliver the fatal blow to their father. In the context of Ovid's narrative, this causation is strongly motivated, of course. We know that Ovid's telling of the myth differed from others in at least this detail-namely, that even under Medea's magical influence Pelias' daughters hesitated (and in fact could not bring themselves) to deliver the coup de grâce (cf. Met. 7.332-8 quid nunc dubitatis inertes?, etc.), whereas in other versions they have no such qualms. 27 This failure compels Medea to do the deed herself. Medea thus stands out against the Peliades as being unconstrained by any filial or familial concern. 28 In this light, the zeugma used to describe the Peliades embodies in its very form the natural order of things that Medea has tried to pervert. In effect, it makes the argument that 'this is the way things ought to be'. In fact, when Medea then takes it upon herself to cut Pelias' throat, this act is expressed in a sort of pseudo-zeugma-cum uerbis guttura abstulit-whose highly irregular syntax seems to capture Medea's deviant morality in linguistic form.
Likewise for (6), si et in urbe et in eadem mente permanent, ea quae merentur exspectent, the upshot of the zeugmatic pairing seems to be to establish a logical connection between remaining in the city (the literal concept) and sticking to one's beliefs (the figurative concept). By this expression Cicero very likely wishes to call to mind a sort of mythic association between physical presence in Rome and seeing through a decision (otherwise exemplified by Livy's story of the centurion whose hic manebimus optime is both a declaration of Rome's defensive suitability and a counterpoint to the Senate's indecisiveness). The argument of the zeugma is that for Catiline's co-conspirators to (literally) remain entails their (figurative) 'remaining', in other words, their acceptance of the consequences of their actions under the law (mereor here has the sense of 'deserve as lawful punishment'). By implication, leaving the city would then constitute a deliberate rejection of legal protections-in clear support of the overall thrust of the speech, which is to convince the people of Rome that Catiline's departure constitutes an act of open rebellion against the state, justifying treating the conspirators as hostes. In (4), de ciuitate decedere quam de sententia maluit, and in (5), de ciuitate maluit quam de sententia demoueri, meanwhile, Cicero deliberately reorients and reinterprets this logical link. Here, the argument of the zeugma is that, when confronted with the impossibility of (figuratively) 'remaining', that is, sticking to his beliefs-in being compelled to swear by illegal legislation-Metellus could not (literally) remain in the city. In this case, Metellus' self-imposed exile then actually implies greater commitment to his beliefs, helping deliver Cicero's characterization of Metellus as an exemplar of aristocratic constancy in the face of populist intimidation.
Something similar can be said of (9), Virgil's oculos dextramque precantem | protendens, as well. In some sense, it is the context of culture that establishes the natural connection between Turnus' (literally) stretching out his hand and his (figuratively) stretching out his eyes towards Aeneas. Fred Naiden has shown that stereotyped gestures of supplication varied between Greek and Roman cultures: Greek supplication typically involved touching the knee and grasping the chin, whereas Roman supplication began with falling to the ground and then extending the hands up toward the addressee. 29 In both cultural contexts, looking directly at the addressee seems also to have played a significant role, at least as indicated by visual sources, which very often represent the suppliant gazing directly (up) into the addressee's eyes. 30 For example, in Figure 1, Odysseus is depicted in the act of supplicating Nausikaa, an act signalled primarily by the directionality of his gaze toward the young girl, who stands behind (or perhaps at the side of) Athena.
The significance of the gaze in Roman supplication appears specially emphasized by the so-called 'Augustus cup' from the Boscoreale treasure (reproduced in Figure 2), where, in the scene depicting Augustus receiving the supplication of the Gallic chieftains, the viewer's focal centre is neither the barbarian on his knees nor Augustus on his stool but the line of sight that extends between them. This line of sight is visually represented by the fold of the lictor's tunic and Augustus' arm, so that the viewer's attention follows the trajectory of the barbarian's gaze. Ovid's zeugma thus nicely embodies the prototypical 'script' for supplication elaborated by Roman society, where looking and gazing are gestures equally constitutive of (as well as, it seems, mutually entailing within) this act.
At the same time, the natural link established by this zeugma between the literal concept (reaching) and the figurative concept (gazing) can be read as part of a persuasive strategy made by the poet in the specific narrative context of the final encounter between Turnus and Aeneas. As Helen Lovatt has claimed, the duel at the end of Book 12 plays out as a sort of 'battle of the gazes' where looking at and harming one's opponent appear to be metaphorically related; as often in epic, the gaze and the sword work as symbolic parallels. 31 Thus, when his sword shatters, leaving him unarmed and helpless, Turnus is simultaneously unable to fixate his gaze; he looks around (circumspicit, 896) and casts his eyes everywhere but on his opponent (he looks for a way out; another weapon; his chariot; his sister Juturna: 913-18). Meanwhile, Aeneas stands menacingly brandishing his spear, 'having marked his target with his eyes' (sortitus fortunam oculis, 920). It is probably significant, then, that, when Turnus has been (literally) wounded, falls to the ground and begs Aeneas for mercy, his sight is finally firmly directed at his opponent (in our zeugma: oculos dextramque precantem | protendens, 930-1), while Aeneas is unable to fixate his eyes upon his (uoluens oculos, 939). In other words, according to the symbolism of this scene, Aeneas is now (metaphorically) weaponless, whereas Turnus is in possession of a metaphorical weapon, namely the gaze. The zeugma, however, has managed to change the terms of the equation-since the gaze here is logically connected not to the sword but to the extended hand of the suppliant. Particularly in light of well-known Virgilian macro-themes that seek to complicate the moral status of Rome in respect to its subject nations, 32 the argument of the zeugma appears to be a kind of recapitulation of Anchises' warning to 'spare defeated peoples' ( parcere subiectis, 6.853), namely that, if unheeded, the claims for mercy of the conquered may be as potentially destructive to Rome as the violence that produced them.
Many of the examples of zeugma I have discussed so far have been taken from verse. This may seem to imply that the naturalizing argument I have claimed is encapsulated in the form of zeugma is particular to poetry, and that in prose this figure instead has just the stylistic function it is normally ascribed in rhetorical handbooks, namely to grab the attention by its unfamiliar and thus striking conceptual associations, as well as by its balanced, almost jingly quality (like what we might today call a 'sound-bite'). In its rhythmic structure, verbal condensation and expressive efficiency, zeugma is certainly well suited to poetic contexts. What is more, I would not want to suggest that every instance of zeugma has to be read instrumentally, as a piece of argumentation. There are cases of zeugma in prose where the figure is likely meant exclusively as a stylistic flourish, such as Cicero's sibi et torquem et cognomen induit 'He [sc. T. Manlius] took for himself both the torque and a surname' (Fin. 2.73) or Phaedrus' conde ferrum et linguam 'Sheath your sword, and your tongue' (5.2). It is difficult to imagine what 'argument' the figure is meant to capture in these contexts, both probably imitative. 33 Usage of zeugma thus appears to cover the spectrum from the more decorative to the more argumentative. None the less, it would be mistaken to assume that prose authors do not also take advantage of the argumentative mechanism afforded by zeugma.
Consider Sallust's pacem an bellum gerens again. The argument of this zeugmawhere the semantically irregular formulation pacem gerere ('wage peace') is 'naturalized' by the conventionality of bellum gerere-is very evidently to elide the two categories. The zeugma therefore recapitulates the point of in incerto, 'since for Jugurtha peace is only a continuation of war'. 34 Or take an expression such as tribuni praefectique cum terrore et armatorum cateruis uolitabant 'the tribunes and prefects hurried about accompanied by terror and bands of armed men' (Tac. Hist. 2.88), where cum terrore must be interpreted figuratively as ablative of accompaniment alongside literal cateruis. In this case, the argument is that the sight of a tribune or prefect in the city with a band of armed guards must inspire fear. In other words, the zeugma states that the relation of armatorum cateruae and terror is one of necessary entailment. Likewise, in C. Curio, summis uestris beneficiis … summo ingenio et prudentia praeditus 'Gaius Curio, endowed both with the greatest talent and prudence and with your greatest favour' (Cic. Man. 68)-where praeditus requires the extended figurative sense of 'honoured with' with beneficiis but its regular literal sense of 'inherently possessed of' with ingenio et prudentia-the zeugma argues, by a kind of implicature, that the former is a natural consequence of the latter.

III. DISCUSSION
In this paper, I have tried to illustrate through the analysis of specific instances of zeugma that this figure of speech, much more than simply drawing an audience's attention to language or heightening its affective appeal-and thus contributing to persuasive discourse as a sort of knock-on effect-can, in certain cases, play an important and perhaps even irreplaceable role in argumentation. In the three cases discussed in this paper-what we can call the 'INTENTIONS AND PLACES' zeugma, the 'SPIRITS AND HANDS' zeugma, and the 'EYES AND HANDS' zeugma-I demonstrated that what authorizes the particular configuration of semantic elements within these constructions is always a metaphorical association that is highly conventionalized in Latin. In other words, the 33 induere cognomen: cf. Plaut. Capt. 820 petroni nomen indunt uerueci sectario. condere linguam: cf. Ov. Met. 3.14 illic purpureis condatur lingua labellis. 34 Lausberg (n. 1), 314 n. 4. semantic pairings that constitute these zeugmatic themes are intelligible and reasonableand thus, in a very broad sense, meaningful-to Latin speakers because of metaphors that are pre-existing and in fact deeply entrenched within their language. Moreover, it is the very conventionality of the metaphors underpinning these pairings that, I suggested, endows zeugma with its argumentative capability. In evoking ordinary, everyday metaphorical associations and indeed in making manifest their ordinariness (by placing literal and figurative side by side), zeugma presents certain symbolic relationships as natural and expected, expectations that authors are then empowered to mobilize within given contexts as an element of persuasion. This is a finding whose significance for the study of Latin literature (or indeed the literature of any language) should not be minimized. In general, it seems difficult to try to separate rhetoric from poetics. 35 Certainly, in the case of Latin authors who were standardly trained in the minutiae of rhetorical theory and writing for an audience whose experience of literature was largely conditioned by oratorical practice, we should actually expect that all the codified techniques of rhetoric should penetrate into their literary meaning construction. 36 But what I have claimed is at stake in these instances of zeugma is much more than the mere encroachment of rhetorical features into literary (and especially poetic) language. I have argued that rhetorical figures-and zeugma in particular-have a role to play in the service of persuasion itself. This sets my claim, and that of the cognitive rhetoric by which it is inspired, against most traditional treatments in which the figures are seen as entirely decoupled from persuasive language and limited to what Jeanne Fahnestock has termed the 'value-added' function, in which they may substitute extraordinary expressions for ordinary ones in order to ornament or defamiliarize a text but do not constitute arguments per se. 37 Yet what do I mean, more specifically, when I claim that a figure such as zeugma can in fact constitute an argument? I mean that the structural configuration involved in a zeugmatic expression (in most cases a single verb used with two objects in somewhat different senses) goes beyond simple punning to represent a formal embodiment of the speaker's meanings and intentions, epitomizing or materializing a certain pattern of thought that the speaker intends the audience to recognize and accept as true. For this reason, Fahnestock has argued that zeugma accomplishes a sort of 'homogenizing' function: 38 by compressing a series of clauses into a corresponding series of object phrases, she sees zeugma as making a determined set of predications identical (so that, for instance, in Pope's 'Great Anna whom three realms obey | Doth sometimes counsel take and sometimes tea' the solemn act of taking counsel and the leisurely one of drinking tea become somehow comparable). However, this suggestion neglects to take into account the fact that very frequently (and perhaps always) zeugmatic pairings depend on symbolic associations that are already highly conventionalized as metaphors in a society's shared conceptual system, and so are capable of representing certain concepts as naturally related, not just incidentally comparable. 39 My claim that metaphor can function in something like this way is hardly new, of course. At least since George Lakoff and Mark Johnson's Metaphors We Live By, metaphors have been seen as conceptual structures, deeply rooted in bodily experience as well as in socio-cultural context, that deliver the understanding of one (usually more abstract) domain directly in terms of another (more concrete) domain. 40 What is new is the claim that the form itself of a zeugma, in evoking entrenched metaphors, also predicates a particular kind of relationship between concepts.
Beyond illustrating that a figure of speech such as zeugma can in fact constitute persuasive speech (rather than merely enhancing persuasion by other means), this analysis has broader implications for scholars working both in classical studies and in cognitive linguistics. To begin with, it suggests that cognitive linguists may need to widen their perspective to consider seriously the role that all kinds of figures may play in shaping thought. While cognitive linguistics has always taken for granted that metaphor and metonymy can operate in this respect-being able to restructure our understanding of particular domains and, especially when organized into systems, even helping to constitute worldviews-very little attention has been given in this perspective to the cognitive functions of other figures, such as irony, personification, hendiadys, or hyperbaton, though these structures equally imply modifications of conceptual relations. It also suggests a need to explore how different figures may interact. Again, metaphor and metonymy have been well studied in this respect, particularly in relation to the question of the grounding of metaphors: some scholars have suggested that conceptual metaphors in fact always depend on more basic metonymical relations. 41 It may also be the case that metaphor, in turn, often underpins other figurative structures, as it evidently does for zeugma.
For classical scholars, this analysis highlights, very generally, the need for further studies of metaphor in Latin, in order to reveal the sorts of conventionalized figurative patterns that organize this language's semantic system. A number of scholars have already made important advances in this area of study, of course: Maurizio Bettini, for example, demonstrated that Latin's vocabulary of temporal relations is organized by twin metaphors of vertical and horizontal linearity. 42 Francisco García Jurado, meanwhile, has shown that orientational metaphors such as 'GOOD IS UP', 'GOOD IS HOT', and so forth, are pervasive in the language of Plautus, 43 and Chiara Fedriani that ontological metaphors such as 'STATES ARE CONTAINERS', 'EVENTS ARE MOVEMENTS', 'EXPERIENCES ARE THINGS POSSESSED' structure much of Latin's grammatical system. 44 My own studies of the metaphor systems converging on Latin speakers' vocabulary of mental activity, of communication and of mistakenness can also be mentioned. 45 Future research will need to look beyond these domains and to trace metaphors as they structure meaning across different levels of linguistic encoding (etymology, lexical and phrasal semantics, grammatical structure, and so forth).
More specifically, it suggests that scholarship on classical literature needs to be attuned to the relationship between conventionalized meaning and imaginative meaning in language. I do not mean that classicists should rehash the debate over which mode of expression (if any) represents the zero-grade of the linguistic code. I mean, instead, that very often the default mode of interpretation in Latin studies is to seek out commonalities of intertextual expression between authors and to treat these commonalities as tokens in some kind of game of literary one-upmanship. Now, there is no denying that imitatio was an important part of Roman practice, and that alluding to, imitating and borrowing from others' works not only particular words or phrases but even whole passages was a central organizing principle of literary production. However, what scholars claim to be instances of imaginative appropriation and adaptation frequently turn out-when viewed against wider linguistic usage-to follow common and highly conventionalized patterns of expression. This implies that literary originality rests not so much in creating new meanings out of whole cloth, as in how authors deliberately elaborate, extend and combine culturally entrenched meanings for emergent contextual or programmatic purposes. 46 So what is creative or imaginative about Ovid's use of the 'SPIRITS AND HANDS' zeugma, for instance, is not the novelty of the semantic pairing-to say that spirits 'fall' is actually commonplace in Latin-but the manner in which he deploys the zeugma within an overall context of meaning making.
By the same token, my analysis suggests that studies of conventionalized metaphorical patterns in Latin must pay attention not only to the grounding of metaphors in human embodied experience and in cultural systems but also to their motivation in determined socio-political environments. 47 The selection of images in metaphorical representation can frequently be accounted for in terms of local culture and global human nature as well as in terms of situated contextual motivations (thus Ricouer's definition of metaphor as a 'planned category mistake'). 48 In fact, it is because the metaphors typically studied by classical linguists tend to be those of a conservative kind-in the sense