Museo Archeologico Nazionale, Naples

Street musicians in a scene from a comedy by the Greek playwright Menander; mosaic by Dioscorides of Samos from the Villa of Cicero, Pompeii, circa 100 BCE

Shortly after the fall of Carthage in 146 BC, the Roman Senate commissioned a Latin translation of a work by a Carthaginian writer. The writer’s name was Mago, and he wrote on agriculture—a subject of deep interest to the Romans, who liked to think of themselves as a nation of sturdy farmers even after they ruled an empire. The work was a substantial one (twenty-eight books, we are told), and it is perhaps no surprise that the translation does not survive, but snippets from it do turn up in later Latin authors writing on the same topic.

The Senate wanted Mago translated because he was thought to have useful information to offer. (As the Roman poet Ovid would later observe, “one can learn even from an enemy.”) It was presumably with similar motives that the ninth-century Abbasid rulers of Baghdad established the bureau known as the “House of Wisdom,” which translated Greek and Syriac works on science, medicine, and philosophy into Arabic. Some of the medical works would later be put into Latin by translators associated with the medical school at Salerno, under the patronage of local rulers. The Arabic translations of Aristotle would also be latinized, in part by scholars working at the polyglot Sicilian court of Frederick II in the thirteenth century.

Another frequent candidate for translation is popular literature: fiction, fables, proverb collections, prophecies, and the like. A classic example is the anonymous piece of ancient historical fiction known as the Greek Alexander Romance, versions of which survive in Latin, Syriac, Armenian, Persian, Arabic, Hebrew, and a half-dozen or more modern languages. Indeed, if we exclude purely documentary texts (laws, treaties, etc.), then these two categories—the useful and the ephemeral—account for the majority of premodern translations.

Denis Feeney’s Beyond Greek: The Beginnings of Latin Literature, studies a case that fits neither category. In 240 BC the Roman magistrates assigned to organize the state festival known as the Ludi Romani (“Roman Games”) commissioned a play-script for performance. This in itself was nothing new. Stage shows derived from Etruscan models had been put on at the festival for over a century. Many Romans would also have seen Greek dramas performed, in Greek, by touring troupes from southern Italy, a region studded with former Greek colonies. What was novel on this occasion was not the play, a Greek tragedy, but the fact that it was presented in Latin. The translator was a native Greek speaker from Tarentum named Livius Andronicus. The name, combining Roman “Livius” with Greek “Andronicus,” suggests he may have been a former slave.

Livius’s first Latin tragedy was followed by others, and he was soon joined by another translator, Gnaeus Naevius. Naevius too was a non-Roman; he came from Campania, where the dominant language was Oscan, an Italian tongue related to but distinct from Latin. Both men translated comedies as well as tragedies. And their efforts extended beyond drama. Livius produced a Latin version of Homer’s Odyssey; Naevius wrote an original Latin epic on the First Punic War of 264–241 BC, in which he himself had taken part.

A generation later both would be eclipsed by a still-greater figure, Quintus Ennius. Like Livius he was a southern Italian, from Calabria. Like Naevius he was trilingual in Greek, Oscan, and Latin; he memorably asserted that because he knew three languages he had three hearts. In addition to translating tragedies he wrote an epic, the Annals, which covered the whole course of Roman history down to his own day. In it he made one crucial innovation. His two predecessors had used a native Italian verse form for their epics, the so-called Saturnian. The rules of this form are still poorly understood, but many surviving lines fall into two unequal halves rather like the English nursery rhyme “The queen was in the parlor,/Eating bread and honey.” Ennius jettisoned the Saturnian, introducing instead a Latin version of the Greek dactylic hexameter, the meter used by Homer and all later Greek epic poets.

None of these works survives intact. In most cases we have only brief quotations in later writers, particularly grammarians who were attracted by their archaic language. The best-preserved of Naevius’s comedies, for instance, is called The Girl from Tarentum. From the surviving twenty or so lines plus the title we can deduce that the girl in question was a courtesan, that she had some connection with two free-spending young men, and that the play was probably not set at Tarentum. Ennius’s survival rate is a bit higher, partly because Cicero admired and quoted from his tragedies, partly because he influenced Virgil and was quoted by ancient Virgilian commentators. But even his longest fragment runs to only eighteen lines, and most are much shorter.


“If someone who speaks three languages is trilingual and someone who speaks two languages is bilingual, what do you call someone who speaks only one language?” So runs a modern joke. The answer, of course, is “an American.” In antiquity the corresponding answer ought to be “a Roman.” Normally, the conquered learn the language of the conqueror, not vice versa. Cleopatra was the first—and last—Greek ruler of Egypt who knew any Egyptian. One would expect the Romans to show similar indifference. And for the most part they did. Sabines and Oscans, Spaniards and Gauls, Britons, Illyrians, and Dacians—all had to learn Latin, and most of their descendants still speak a language recognizably related to it. Yet the Romans did learn Greek. Indeed, they ran the eastern half of their empire in it, from Egypt all the way to the Black Sea, as well as Greece itself.

This striking phenomenon has a traditional explanation. The Romans, a younger, cruder culture, could not help but be impressed by the Greeks’ cultural achievements even as they trounced them on the battlefield, first in southern Italy and then in the East. “Once made captive,” as Horace famously put it, “Greece captivated her captor.” And the Roman obsession with things Greek has apparent parallels in other cultures: the enormous influence of Chinese culture on Heian-era Japan, or nineteenth-century Russian aristocrats’ fascination with French. How natural, then, that the first recognizable works of Latin literature should be translations from Greek.

Yet as Feeney points out, translating another culture’s high-grade literary texts is almost unexampled in antiquity. No other people in the ancient world did this: not the Egyptians, not the Phoenicians or their colonists at Carthage, not any of the other Italian peoples, the Oscans, Umbrians, or Etruscans. And certainly not the Greeks. The one apparent exception, the Greek translation of the Hebrew Bible known as the Septuagint, was carried out by Alexandrian Jews for their own Greek-speaking community. (The legend that it was commissioned by one of the Ptolemies has no foundation.) The nearest real parallel seems to be the Akkadian translations of the Sumerian Gilgamesh epic, which were certainly not known to Livius Andronicus.

If we accept that the Romans would have developed “literature” sooner or later (and as Feeney notes, there is nothing inevitable about this), Roman writers might have taken their own path, developing their own conventions and genres, as Egyptians and Jews did. Why didn’t they? Was Greek literature simply too appealing, too powerful to be ignored? Perhaps. But in that case why did the Romans not simply write in Greek?

This, after all, was a path followed by a number of non-Greek writers, including the Egyptian chronicler Manetho, the Jewish historian Josephus, and the Alexandrian Ezekiel, who composed an account of the Exodus in the form of a Greek tragedy. Indeed, the first Roman historian, Fabius Pictor, did write in Greek, as did one or two of his immediate successors. Greek was widely spoken on the Italian peninsula, including at Rome itself. Far from bursting upon the Romans in the third century, Greek culture had always been there. In the late Republic, elite Romans were effectively bilingual; one can imagine Catullus writing poetry in Greek just as the aristocrats of Heian Japan did in Chinese.

The first Roman translations are unusual in other respects too. Scholars of translation distinguish between what they call L1 translation (into one’s native tongue) and L2 translation (into another language, usually one less perfectly controlled). Almost all modern literary translation is L1, from Gavin Douglas’s sixteenth-century Eneados to Don Bartlett’s twenty-first-century Knausgaard. L2 is what produces the assembly instructions for your new baby stroller. But Livius and his colleagues are literary L2 translators, moving from their first or second language (Greek) to their second or third (Latin). This is not normal.

In thinking about these issues Feeney acknowledges a considerable debt to the work of David Bellos. Bellos is an accomplished translator from French (notably of the novelist Georges Perec); his Is That a Fish in Your Ear? is a shrewd and delightful book that explodes many myths about translation while putting forward much that is new and interesting.1 As Bellos emphasizes, the relationship between source and target language is almost never one of equality. Rather, translation involves movement between a language of lower status (defined by prestige, popularity, or both) to one that is comparatively higher. To use Bellos’s terms (and typography), translators translate either UP or DOWN. Translation UP generally tries to elide the foreignness of the source—to produce a text as much like one composed in the target language as possible. Translation DOWN tends to preserve linguistic features of the source language, since these carry cachet in and of themselves.


Livius and Naevius translate UP; that is to say, they Romanize the texts they translate; in so doing they mark Latin as UP with regard to Greek. Where Homer had invoked the Muse, Livius’s Odusseia calls on an Italian deity, one of a group of water nymphs known as the Camenae. His Nausicaa rides not in a wagon, as Homer’s does, but in a two-wheeled Roman vehicle, a carpentum. In Ennius’s epic (though not his tragedies) the balance begins to shift: he discards the Saturnian for the Greek hexameter, and the Camenae for the Muses. Paradoxically, as L1 translators take up the baton, the “Greekness” of Roman translations increases: “There is far less Grecism in Livius and Naevius,” Feeney writes, “than there is in Catullus or Virgil.” For Catullus, Latin was DOWN compared to Greek. Yet for Livius it had been UP. What does that imply?

It is tempting to regard the early dramatic translations as a natural response to audience demand. Like spectators at the English National Opera, Roman viewers wanted to see the show and understand the words. But the notion that the first play in Latin just popped up like a mushroom is difficult to reconcile with what we know of Roman society and Roman theater. Throughout its history the Roman state was concerned to monitor and control public displays of all kinds, and theater in particular. No permanent theater was allowed in Rome itself until the first century BC; seating restrictions were designed to keep the social classes apart.

The Ludi Romani were a state festival supervised by magistrates; the stage performances would have been closer to a Super Bowl half-time show than to street theater. The staging of a Greek tragedy, in Latin, as part of the proceedings can hardly have been a spur-of-the-moment initiative. (“Hey, why don’t we put on the show right here in the Forum!”) It must have been a conscious and carefully made decision.

What Feeney provocatively dubs “the translation project,” then, is not a natural outgrowth of Roman cultural dependence but a deliberate act of almost breathtaking audacity. By producing Greek tragedy in Latin, the Romans were asserting parity with the Hellenistic Greek kingdoms that had long treated drama, like triumphal parades and religious ceremonies, as a form of politics by other means. The echo of the Manhattan Project in Feeney’s phrasing is implicit but clearly deliberate: Latin tragedy as A-bomb, with Livius and Naevius as Fermi and Oppenheimer.

It is worth noting here that the Manhattan Project was not originally called the Manhattan Project. In early paperwork it is referred to by the innocuous phrase “Development of Substitute Materials.” Feeney, I suspect, would find this a pleasing coincidence. As Bellos observes, people like to assert that “a translation is no substitute for the original,” but of course that is precisely what a translation is: a substitute for the original. A Greek visitor to the Ludi Romani—say, an ambassador from one of the Hellenistic rulers—might have found himself nervously wondering what other kinds of substitution the Romans proposed to make.

Menander; Roman fresco, Pompeii, first century AD

It is significant, on this reading, that the translation project made its debut in 240 BC, the year after the end of the First Punic War—or as Feeney suggestively refers to it, “the Great War.”2 This is a moment when Rome had consolidated its control of Italy, extended its dominion to Sicily, and was beginning to look further afield—beyond Carthage, even, to the Greek kingdoms in the east. The translation project, then, is merely one component of a larger program. Following Feeney’s lead, we might call it “the domination initiative.” Feeney shies away from asserting “that the Senate from the start had some kind of grand design of appropriating Attic drama,” but his argument seems to point to something very like that.

One might wonder whether such a reading credits “the Romans” with too developed a capacity for collective strategizing. This question in turn mirrors a larger debate over the nature of Roman expansion. Was Rome’s empire the result of a conscious grand strategy deliberately pursued over several generations? Or did the Romans acquire it (as the British are alleged to have done) in “a fit of absentmindedness,” as a byproduct of status jockeying between individual aristocrats?

Feeney reads the Ludi Romani of 240 as part of a high-level (and high-stakes) dialogue between Rome and other Mediterranean powers. And perhaps they were. But Roman festivals were also vehicles for one-upsmanship by individual magistrates. When Cicero went off to govern Cilicia in the late 50s BC he was bombarded with requests from his younger friend Marcus Caelius, who was running for aedile (a lower-tier office something like a commissioner of public works) and putting on games as part of his campaign. He thought it would be nice to have panthers. Could Cicero get him panthers? Cicero was dubious; there was a panther shortage, but he would do his best. Were Livius’s verses perhaps the third-century equivalent of Caelius’s panthers, the “something familiar, something peculiar” with which every magistrate aimed to wow the public? Feeney has some gentle fun with such reconstructions:

It is easy enough to write a novel in which Livius Andronicus is commissioned for the special peace games because his former master, or his patron, L. Livius Salinator, is on the committee planning the games. Salinator could remark, “I have a Graeculus [“little Greek”] at home who might be able to do something special for us”; and after the performance was over we can imagine the nobiles…saying to each other, “That was very worthwhile; let’s do it again next year.”

In reality there is nothing inherently implausible in such a scenario. We simply do not know how the mechanics of Livius’s commission worked, or who exactly commissioned him. But regardless of whether Feeney is right that “the initial impulse to transform the Ludi Romani was state policy,” he performs a valuable service by defamiliarizing what had hitherto gone unquestioned. The birth of Roman literature now begins to appear as what he calls it: “One of the strangest and most unlikely events in Mediterranean history.”

Among the idées reçues skewered by David Bellos is the old saw that “poetry is what gets lost in translation.” The saying is often attributed to Robert Frost, but as Bellos notes, the attribution is as dubious as the idea itself. A translation is an assemblage of words, and as such it can contain as much or as little poetry as any other such assemblage. The Japanese even have a word (chōyaku, roughly “hypertranslation”) to designate a version that deliberately improves on the original. The Romans can make some claim to have done this too.

Our fullest insight into Roman adaptation of Greek drama is provided by the comedies of Plautus, working in the late third and early second centuries BC, a generation or so after Livius, and those of Terence (a generation later still). Both based their work on the so-called new comedy of fourth-century Athens whose best-known playwright was Menander. Twenty of Plautus’s plays survive (along with part of a twenty-first). Here we have more than fragments, but the loss of most of the Greek originals had made it hard to evaluate them as translations. Some help came in 1907 when the sands of Egypt yielded up a papyrus codex with remains of three plays by Menander, though none of them turned out to underlie an extant Latin play.

The year 1922, which brought us both Ulysses and The Waste Land, also produced the original German edition of Eduard Fraenkel’s Plautine Elements in Plautus, a work as epoch-making in its own field as Joyce’s or Eliot’s. The title is consciously paradoxical (as if one were to write a book on Milton’s contribution to Paradise Lost or Shaw’s influence on Pygmalion). Yet it is also perfectly descriptive. Fraenkel’s study attempted to work out, by analogy from Terence and what little Menander was then available, what Plautus must have done to his Greek models. Since the original publication, more Menander has come to light. Notably, the late 1960s brought a fragment of his Dis Exapaton (The Double Deceiver), which underlies part of Plautus’s Two Girls Named Bacchis, thus permitting us for the first time to compare a substantial chunk of Plautus with his model.

The new finds have largely confirmed Fraenkel’s insights. Plautus retains the plot of his originals but adds references to Roman institutions, puns that only work in Latin, slapstick humor, and the elaborate aria-monologues known as cantica. Another Plautine contribution is the stock character of the clever slave, who plays much the same role as the “magical Negro” of contemporary American movies—a socially marginal figure who advances the interests of a more privileged but less worldly-wise patron. (In Roman comedy this is the Woosterish young man around whose frustrated amours the plot typically revolves.) It may be that Livius and Naevius anticipated some of these elements. Certainly they prepared the way for them, and for a progeny that runs from The Comedy of Errors to Noises Off. Whether Plautus’s plays are “better” or “worse” than Menander’s is something one can argue about. Either way, they were certainly not “just translations.”

Perhaps the greatest achievement of the Roman translation project was the imaginative space in which Plautus’s characters move, an area nominally Greek, yet liable to turn briefly Roman at any moment. Roman writers continued to exploit this device well into the imperial period. Virgil’s Eclogues are set in a Greek province, Arcadia, whose inhabitants experience the Italian land confiscations of the civil war period. Seneca’s Thyestes presents, as one scholar puts it, “a strange hybrid world” where Roman household gods protect a Greek palace, and Parthian horsemen mass on the borders of the Peloponnese. Apuleius’s Golden Ass, written in the mid-to-late second century AD, offers us a confessedly “Grecistic” tale in which Thessalian stableboys quote Cicero and the Delphic oracle prophesies in Latin hexameters. If all Russian novelists emerged from under Gogol’s “Overcoat” and all American writing from Huckleberry Finn, then all of Roman literature descends from Livius’s Ulixes (as his Odysseus was named). Homer had called his hero polytropos, “full of twists and turns.” Livius made him uersutus, which means much the same—but, significantly, can also mean “translated.”