|By V. Joshua Adams
|March 11, 2021
In the 1997 film directed and co-written by Roberto Benigni, La vita è bella (Life is Beautiful), the main characters, including Guido (played by Benigni himself) and his five-year-old son Giosué, are deported from Tuscany to a Nazi concentration camp. To protect his son from the grim truth, Guido improvises an elaborate fantasy: Giosué must follow his father’s instructions in order to win points for a game. In one of the film’s iconic scenes, Guido volunteers to serve as a translator of the camp’s rules, using the opportunity to transform the remarks of a German officer into the rules of the game he has invented for his son:
Guido takes his place next to the corporal, who immediately starts yelling loudly at the prisoners.
Corporal (in German): Attention! I am only going to say this once.
He looks at Guido, waiting for a translation.
Guido: (in the same military tone) Okay, the game begins! If you’re here, you’re here, if you’re not, you’re not!
Corporal (in German): You are here for one reason, and one reason only!
Guido (in Italian): The first one to get a thousand points wins a real tank!
Corporal (in German): To work!
Guido: Lucky man!
The prisoners are trying to understand what the Corporal and Guido are talking about, but only Giosué gets it. He is standing very still, eyes popping.
The corporal points to the yard outside.
Corporal (in German): Any sabotage is punishable by death, sentence carried out right here in the yard, by machine gun to the back!
He points to his own back.
Guido: Scores will be announced every morning on the loudspeakers outside! Whoever is last has to wear a sign that says “jackass” — here on his back!
Imitating the Corporal’s gesture, he points to his own back.
Corporal (in German): You are privileged to work for our great Germany, building our great empire!
Guido: We play the mean guys, the ones who yell. Whoever’s scared loses points!
There is much to say about this scene, but I want to ask a very basic question: is Guido translating? I’m guessing most of us are going to say, “No.” Of course, it looks like Guido is doing what translators do, namely, making one language comprehensible in terms of another. But Guido’s not actually doing this, since he doesn’t understand the statements he is supposed to be translating. He lacks a grasp of the semantics and the syntax of German that would allow him to understand, and thus to translate, the rules of the camp. It’s true that, in the scene, he mimics the intonations of the German officer’s voice and the gestures of his delivery (like pointing at his back and, later, counting to three on his fingers). But these of course only emphasize the degree to which Guido does not know what the officer is actually saying and, as Sander Gilman once pointed out, work to comic effect by highlighting the degree to which the Germans are ignorant of how their “rules” are being misunderstood.
Then again, even if we agreed that Guido is not translating from German, we still might be tempted to say that he is “interpreting” the remarks of the German officer. After all, there is much that Guido does understand about his situation. His invention of the game for Giosué strikes us as poignant (rather than, say, irresponsible) insofar as, on some level, he and the other adult prisoners can assuredly surmise that the “mean guys, the ones who yell,” are not playing a game at all. True, Guido can’t explain or justify any of his choices with respect to the meaning of the corporal’s speech. But, in this particular context, he doesn’t need to. There is only one person for whom he needs to make one language comprehensible in terms of another: his son. What exactly would prevent us from calling this an interpretation and, because it involves more than one language, an instance of translation? What’s the difference between translation and interpretation and making it up as you go?
This question bears directly on Lawrence Venuti’s new book, Contra Instrumentalism: A Translation Polemic. In the book, claims are made for a new “model” of translation, one premised on the fundamentally interpretive dimension of the translation process. The potential arrival of a new model of translation is a significant event. As George Steiner reminded us in his 1975 study After Babel: Aspects of Language and Translation, “despite [a] rich history, and despite the caliber of those who have written about the art and theory of translation, the number of original, significant ideas in the subject remains meager.” Steiner himself did not so much add to those ideas as collect and arrange them, in order to suggest that the ensemble was not peripheral but central to the larger project of understanding “the operations of language itself.” Venuti’s project is even more ambitious: he suggests that once we adopt his new model of translation, we will be able to see translation as central not only to the arts and the human sciences but also to modern life more generally. All this will require is a wholesale rejection of everything we thought we knew about how translation should be done.
Contra Instrumentalism is, as the title indicates, a polemic, but it is also supposed to be more than this. Venuti claims it is an example of what Gilles Deleuze and Félix Guattari called a “desiring-machine” or “machinic assemblage.” The book means to work on the reader in such a way that it produces “the will to critique a model that has been deeply entrenched in thinking about translation for so long as to be unconscious, knee-jerk, rote.” In order to critique and replace such longstanding assumptions about translation, Venuti wagers that we need more than what he calls “coolly detached reasoning,” but rather “the release and redirection of desire.” No doubt desire plays an important role in our willingness to engage in critique in the first place, and to sustain it. Venuti trusts that his readers desire what he desires, which is to change the status of translation studies by changing the way we think about translation. But the logic of his critique of existing thought about translation is not only responsible to desire but also to reason. And it is here, in the space of its reasoning, that Contra Instrumentalism runs into some problems.
Venuti’s preferred model of translation does look both new and radical. Indeed, it appears so new and radical that it would make us question even our most instinctive, common-sense thinking about translation — like our instinctive, common-sense rejection of Guido as a translator. But in reading Contra Instrumentalism, that is not actually what happens; the radical promise of the book remains unfulfilled. The reason for this, I will suggest, is twofold. First, Venuti mischaracterizes the differences between his model of translation, which he calls “hermeneutic,” and its alternative, which he calls “instrumental.” Initially presenting these models as if they represent an either/or choice, he eventually concedes that they are not as opposed as they first appear. Second, Venuti mischaracterizes the relationship between translations and their sources. He writes as if translations replace sources when they actually represent them. Like the categorical distinction between hermeneutic and instrumental models of translation, this is a false picture, one that Venuti’s discussion of specific translations dispels. The radical theory of the book — which turns out to be less radical, and less coherent, than it appears — is also contradicted by its practice. This does not negate the necessity of Venuti’s polemic per se, but it does force us to reconsider where and how it might actually be profitably pursued. The real contribution of Contra Instrumentalism does not come from its theorizing but from its practical criticism. Here Venuti succeeds in pointing a way forward for translation studies, albeit in different terms than he might like.
As the title indicates, Contra Instrumentalism struggles against a prevailing view of translation, which Venuti conceives of “as the reproduction or the transfer of an invariant that is contained in or caused by the source text, an invariant form, meaning, or effect.” Against this “instrumentalist” model — which, per Venuti, saturates both academic and popular discussion of translation and is responsible for the institutional marginalization of translation as a practice and as a subject of study — he proposes an alternative, “hermeneutic” model, which “conceives of translation as an interpretive act that inevitably varies source-text form, meaning, and effect according to intelligibilities and interests in the receiving culture.” The payoff of favoring the hermeneutic model is the abolition of what Venuti once called, many years ago now, “the translator’s invisibility.” Adopting it will force us to pay more attention to translation, allowing us to reverse trends in which translation is overlooked in literary or academic practice, and to resist dominant political and economic paradigms (since both politics and economics, in an international context, depend on translation in various ways).
Venuti’s polemic depends on the distinction between these two models of translation, which he (initially, anyway) presents as alternatives. The sharpest distinction between them is the presence or absence of an “invariant.” The instrumentalist, according to Venuti, believes that something about the source text does not change in the translation process, and this is what enables the translator to “reproduce” whatever the invariant is; the hermeneut believes that translation varies the source, and this is what enables (or perhaps even requires) the translator to produce something that responds to his or her own context. In the case of the Benigni example, if I think my job as a translator is to bring the form, meaning, or effect of the German officer’s remarks into Italian, I am an instrumentalist, since I have assumed the persistence of that form, meaning, or effect across languages. If I think my job as a translator requires me to fundamentally change the German officer’s remarks in order to make them comprehensible in Italian, I am a hermeneut, since I have assumed that I will need to vary the source-text form, meaning, or effect in order to create my translation.
The term “invariant” is uncommon in translation studies and in the history of thinking about translation. Venuti himself gives no definition or genealogy of the term. My sense is that it comes to him from Deleuze and Guattari, who inherited it from Lacanian psychoanalysis and Levi-Straussian anthropology, both of which used the term to describe unchanging structures of the mind. Certainly those are the contexts that Deleuze and Guattari themselves invoke in Anti-Oedipus when they write that “We in no way claim to be taking up an endeavor such as Malinowski’s, showing that the figures vary according to the social form…. We even believe what we are told when Oedipus is presented as a kind of invariant [une sorte d’invariant]. But the question is altogether different: is there an equivalence between the productions of the unconscious and this invariant — between the desiring-machines and the Oedipal structure? Or rather does the invariant merely express the history of a long mistake … the strain of an endless repression?”
It is important to note that, in this context, and regardless of your sympathies in the structuralist / poststructuralist debate, invariants are not merely “forms, meanings, or effects.” Rather, they are the deep structures that produce forms, meanings, and effects, the conditions of possibility such that we can be said to encounter forms, meanings, and effects at all. Invariants like the incest taboo or the Oedipal story (or the Mirror Stage, etc.) are invariants insofar as, to put it in Saussurean terms, they are langue and not parole. (Deleuze and Guattari, in their call to “schizophrenize” rather than “oedipalize” the unconscious, ratify this distinction by trying to change the langue en route to an encounter with their own invariant, “desiring-production.”) Venuti treats invariants rather differently than these French forebears do. In his hands, they are instances of parole rather than langue, discrete phenomena or practices that we have the capacity to avoid or refuse if we adopt the right interpretive or translational stance. Rather than seeing our job as one of reproduction of the invariant, we should see ourselves as producing something new, and he thinks we are in the position to do this, provided we follow his “hermeneutic” model. A translator who adopts such a model does not reproduce a meaning, form, or effect but “turns a source text into a translation by applying interpretants, factors that are formal (such as a concept of equivalence or a concept of style) and thematic (such as an interpretation of the source text presented elsewhere in the commentary or an ideology in the sense of an ensemble of values, beliefs, and representations affiliated with particular social groups).”
However modest and technical this language might appear compared to something like Deleuze and Guattari’s dramatic call to “shatter the iron collar of Oedipus and rediscover everywhere the force of desiring-production,” ask yourself: what exactly would stop someone from describing Guido’s approach to translating as turning a source text into a translation via the application of interpretants? Even though he has no grasp of the meaning of the German words or phrases in the ordinary sense, Guido’s responses do resort to, and their humor depends on, various kinds of stylistic equivalence, including clausal length and vocal stress. And certainly his own awareness of the German invasion that occasioned his deportation, his experiences living under Italian fascism and anti-Semitism before the German invasion, and his role as a father and husband would potentially count as what Venuti calls “thematic” interpretants in the sense of values, beliefs, and representations — not embedded in the German officer’s remarks, exactly, but surrounding them.
Taking seriously the idea that a translation not only interprets but also changes a source, as Venuti suggests we should, is a radical idea. And what we might call a resolute reading of the hermeneutic model invites us to draw the conclusion that, because translation can change a source, Guido is translating or interpreting literally as well as metaphorically. After all, why should it matter if he understands the original if he is destined to change it? Here we recall that “hermeneutic” in Venuti’s hands does not invoke the tradition of philosophical hermeneutics that stretches back through Gadamer and Heidegger to Dilthey, Schleiermacher, and Herder, for whom the claim that Guido is translating would be absurd. (Indeed, in a remarkable display of chutzpah, Venuti sends philosophical hermeneutics packing as a crypto- instrumentalism in the first few pages of his book by summarily dismissing Heidegger’s “The Anaximander Fragment” — as if that example could stand in for philosophical hermeneutics in general.) The difference between hermeneutics proper and Venuti’s hermeneutics is not the fact that “no translation can be understood as providing direct or unmediated access to its source text”— this much has been clear since Herder and Schleiermacher. Nor is it even the idea that “the necessary mediation of interpretants allows any text to support multiple and conflicting interpretations as well as to give rise to many different translations.” That news is old as scripture. Rather, the novelty of Venuti’s position is that, amidst various mediations and conflicting translations, a confluence of contexts, and not a reciprocal relation of actions and contexts, ultimately determines the interpretation. There is no hermeneutic circle or spiral but rather a one-way movement from context(s) to text. He writes: “Interpretation potentially releases an endless semiosis that is delimited by an interpretive occasion, an institutional site, a conjuncture of cultural forms and practices, and a historical moment — by, in other words, changing interrelated and mutually determining contexts of interpretation that can each lead to different translations of the same source text, a condition of language that Derrida calls iterability.” So long as we find the right interpretive occasion, the right institutional site, the right conjuncture of cultural forms and practices, the right historical moment, then Guido is just as much of a translator as anyone else, and maybe even a more ethical one — for the purposes of preserving the innocence of a child, say.
Here we have the translator liberated from the iron collar of the source text. The benefit of this model is that anything can be a translation (if it satisfies the right conventions). The problem with this model is that anything can be a translation (if it satisfies the right conventions). Close attention to Venuti’s prose reveals that he is sensitive to how such an aporia would render the idea of translation itself meaningless, at least “potentially” — a justified fear, which also reveals that, for all of his apparent polemicizing, he is not quite as resolute a proponent of the hermeneutic model as he appears. In the remark above, for example, the Derridean translators are generating iterations of the same source text, so something “invariant” or, at the very least, persistent — and, by the way, the difference between “invariant” and “persistent” is crucially unexplored in the book — structures the practice of the hermeneut. Moreover, it’s not just the idea of a source that persists but also what we are supposed to do with it. Clearly aware of the potential conflict between the translator’s obligation to transform a text for his audience and his responsibilities to the source author’s meaning — the dilemma between what he has famously called “domesticating” and “foreignizing” translation — Venuti writes, “a translation does not simply assimilate the source text to what is intelligible and interesting to receptors. By maintaining a semantic correspondence and stylistic approximation, a translation can provide a basis for various accounts of the source text, including plot summaries and character analyses, summaries of philosophical arguments … explanations of conceptual terms, descriptions of lexical and syntactic features as well as their coalescence into a distinctive style.” This careful and qualified list of the many different achievements of semantic correspondence and stylistic approximation, however helpful, ignores the obvious “invariants” of “semantic correspondence and stylistic approximation” themselves that preside over the translation process. Venuti is certainly correct that translation is fundamentally interpretive. But the conclusion that he draws from this premise — that because it is interpretive it is free of “invariance” — is mistaken, because it ignores the invariances that produce the category of translation itself.
To return to the example of La vita è bella, the very fact that semantic correspondence is a necessary condition for translation (though perhaps not always a sufficient one) enables us to say, rightly, that Guido is not translating. The German words are not actually words for Guido. They are sounds that resemble words, signifiers without signifieds. One can surely respond to sounds by producing words that resemble them sonically. But while purely homophonic translation produces art, maybe even great art, it cannot produce understanding. Given that semantic correspondence is necessary for translation regardless of which model we choose, it seems that the distinction between the instrumental model and the hermeneutic one is one of degree and not kind. Both the instrumentalist and the hermeneut apply interpretants in translating, but in order to apply the interpretants they need to have some understanding of the source. Without this fundamental understanding, anything goes, so far as our contexts permit, leaving us with not only the possibility of Guido as a translator but also the even more disturbing possibility of having no way of judging that something was or wasn’t a translation, since it could always be a case of our lacking the right context.
That the main distinction of Contra Instrumentalism might be one of degree and not of kind is not a liability in itself. But it restricts the extent to which the interventions of Contra Instrumentalism can be called “radical,” and thus it poses problems for the way Venuti frames his polemical claim and for his specific remarks on the two models. When he writes of the hermeneutic model that “any correspondence or approximation [produced] thus coincides with a radical transformation,” the word “radical” can mean “major” or “important” or “comprehensive,” etc. But it can’t mean radical in the sense of “fundamental” or “categorical” because differences of degree are not radical in the way that differences in kind are. In translation, correspondence or approximation don’t just “coincide” with transformation but shape it.
Why does Venuti commit to a rhetoric of radicalism about translation when his argument is not really radical? In part, he conceives of the study of translation as a means of pursuing radical social change, and his sense of political responsibility informs and suffuses his scholarship. But, however committed Venuti is politically, I think this explanation is too simple. In reality, what I’m calling a rhetoric of radicalism is produced by a different commitment that is not primarily political but ontological. It is a commitment to the idea that when you translate a text you actually do something to it. Not with it, mind you, but to it. Here, for example, Venuti explains why a translation will never enable a non-native reader to respond to a source text in the way that a native speaker might: “…a text is a complex artifact that sustains meanings, values, and functions specific to its originary language and culture, and when translated this complexity is displaced by the creation of another text that comes to sustain meanings, values, and functions specific to a different language and culture” (my italics). In this view, translation takes the place of the original and, in its stead, sustains different meanings, values, and functions. Thus it looks like it makes sense to think of the transformation as “radical” because one text and language has been literally substituted for another.
Wittgenstein famously remarked in The Philosophical Investigations that “A picture held us captive. And we could not get outside it, for it lay in our language and language seemed to repeat it to us inexorably.” Venuti’s claim that translations displace their sources is one such picture. To be sure, he is hardly the only translator or translation theorist to think this way. (No doubt I’ve done so myself.) And it’s clear why the misunderstanding persists. When we talk casually about translating source texts, it does sound as if we are doing something directly to those texts. When we talk about whether a translation captures or distorts the meaning of an utterance, it does sound as if our translation is, as it were, in contact with the meaning of the source. But what’s really happening at the most basic level when we translate is that we are making a representation of a text. To do this, we have to interpret, to be sure. But, just as a landscape painting or a portrait is distinct from the landscape or the person, and an actor distinct from the role she plays, a translation is distinct from its source. Our translation denotes or refers to a source; in doing so, it can also express and exemplify it. But it does not, merely through the act of translation, displace or replace it. It would be curious to say seriously that, in painting a landscape or a person, we had actually done something to that landscape or that person or that, in performing a role, we had actually done something to the script — apart from, in all of these cases, saying that we had interpreted and represented it.
And yet, just as with landscapes and portraits and performances, the sources of a translation are often absent for comparison. Most translations are not published along with their sources. (Poetry, which is sometimes translated in facing-page editions, is an occasional exception here.) Thus it might reasonably appear that the translation is taking the place of the source within the target language, since the source text is absent. And it might appear to do this in an even more dramatic way, say, than the other examples I’ve listed, since the source is not only absent but also largely inaccessible to readers unless they happen to know the source language.
Daunting though practical barriers to access may be, they are not the same thing as a substitution of one thing for another. No translation actually takes the place of the source text. It is true that, practically speaking, translations primarily circulate in the target language and shape how sources are perceived in the target language. These perceptions can affect and distort larger judgments about texts, authors, source languages, and cultures. Translations can become source texts for other translations or for interpretations. This can occlude their origins as translations, leading to false senses of linguistic stability and cultural chauvinism; it can also be a boon for writers, as Pound pointed out: “A great age of literature is perhaps always a great age of translations. Or follows it.” These are all important practical issues that demand attention, criticism, and often intervention in the form of re-translation. The whole long history of the critique of representation and all of its tools can and should be brought to bear on translations. But the seriousness of these practical issues does not oblige us to make false theoretical claims about what translations do to sources. Indeed, if such theoretical claims were true, and translations did wholly displace source texts, there would be no basis for criticizing them (or praising them) as translations at all.
The ontological mistake of Contra Instrumentalism underwrites the rhetorical one. Both are fueled by an admirable desire to get professional intellectuals to pay more attention to translation, but they produce a blinkered account of what it might mean to do this. As a result, even as he concedes that, while the “models may seem to form a binary opposition … on further scrutiny any such opposition collapses to reveal the primacy of interpretation,” Venuti focuses exclusively on the way in which adherence to “formal and semantic” invariance, which he identifies with instrumentalism, might “[preempt] thinking about translation as an interpretive act.” What he neglects to add is that, given the unity of the models, a categorical adherence to formal and semantic variance would lead us away from thinking about translation as an interpretive act, too. Pursued without recourse to some non-varying standard, translation would float free of the source. In order to interpret, you need something to interpret, and Venuti consistently downplays how the source will always, to some degree, serve as a constraint on the translator.
Perhaps we should forgive this failure of even-handedness in a polemical book. But Venuti’s rhetorical commitment to a position that is more radical than the one he holds — indeed, more radical than any translator could hold — leads him into making uncharitable and ultimately incoherent criticisms of many other translators. These criticisms aren’t merely unproductive and distracting. They show how his polemic is more tendentious than it needs to be. Consider, for example, his take on Emily Wilson, who recently produced a widely praised translation of the Odyssey. Venuti quotes an interview with Wilson, in which she says, quite reasonably, “all translations are interpretations.” But then he notices something else she says: “I want to be super responsible about my relationship to the Greek text. I want to be saying, after multiple different revisions: This is the best I can get toward the truth.” Venuti accuses Wilson of contradicting herself by, effectively, embracing both models of translation at once. This is unjust, not only because Venuti implies that Wilson is using truth as “single and definitive” when she does not say so (the truth of the matter might be complicated, or ambiguous, and, in any event she is only positioned “toward” it) but also because, as he himself acknowledges, there is no fundamental contradiction between the models between which she oscillates in the first place! Venuti understands this, which is why, in juxtaposing himself to Wilson, he resorts to a rhetorical sleight of hand: “The fact is that the hermeneutic model, in understanding translation as variable interpretation, renders inadequate any appeal to the source text as the sole justification for a particular translation” (my emphasis). Of course, Wilson does not say that the source text is anything like the sole justification for a particular translation. She is implying that it is necessary — that without it, her translation could not be true to its source — but she is hardly saying that it is sufficient.
Why make such an obviously erroneous attribution? Venuti’s account of translation, having assumed that translations replace their source texts, cannot see a role for those sources in accounting for the success, failure, or even the very rationale, of a translation. Of Wilson’s reliance on the Greek text, Venuti writes,
The substantiating force of such an appeal must always give way to the relation that the translation establishes to the conditions that figure in its production and reception, conditions that are linguistic, cultural, and social. With a text that has been retranslated as many times as the Odyssey, any justification for yet another version should distinguish it against previous interpretations and translations, ultimately considering their position in the hierarchy of values, beliefs, and social representations in the receiving situation. This hierarchy, a ranking according to cultural authority or prestige, matters for the viability of the interpretation inscribed in the source text, since it will inform readers’ responses regardless of whether the translator takes it into account. Those responses will of course vary according to diverse factors, personal as well as transindividual, but foremost among them is likely to be whether and how a retranslation conforms to or challenges dominant interpretations of the source text. By asserting the Homeric text is the container of truth, Wilson has effectively suppressed these considerations.
This passage gives readers a good sense of the strengths and weaknesses of Contra Instrumentalism as a whole. Essentially, Venuti argues that a translator’s work should not be evaluated against the source, as Wilson claims she is doing, but against the existing norms of expression and evaluation within the target culture. Even if you have an account of why your translation is the right one, or the best one, what counts as “right” or “best” — what Venuti calls, wearily, “the justification of yet another version” — is ultimately determined by what the target culture will accept. Let me say that this is certainly a non-trivial point, and the acclaim for Wilson’s translation may have something to do with a tendency to linguistically “domesticate” Homer. But despite its importance, the point Venuti makes also presents us with an incomplete picture of the situation. When it comes to translation, the hierarchy of values, beliefs, and social representations in the receiving situation is important, but it is not the only thing that matters. There is also the matter of the translator’s understanding of the source text, and the source itself, both of which function as limits on what she can do. These boundaries are neither strict nor absolute, and many of us may lack the qualifications to contest their being drawn in a particular way, but none of this would imply that the boundaries are irrelevant, that appeals to the text “must always give way” to something else. The fact that a source comes to us mediated does not mean that the source does not count.
When Contra Instrumentalism stops polemicizing and theorizing, and starts engaging in practical criticism of translation, it becomes an altogether more provocative and profound book. Examples abound in Venuti’s discussion of proverbs, but the strength of his practical criticism comes through best in his discussion of film, “The Trouble with Subtitles.” The chapter shows how subtitles, translations that so often impoverish and flatten the meaning of cinematic dialogue, can also enrich and complicate it in ways that remain largely unexplored by both subtitlers and critics. Constraints that source texts and semantic equivalence place on translators are not debilitating, either to creativity or criticism. In fact, these constraints can be enabling.
Here is one of many compelling examples that Venuti provides. In the 1977 film Annie Hall, Woody Allen’s character, Alvy Singer, is complaining to his friend Rob over lunch about hearing what he thinks is an anti-Semitic slur spoken to him by another man, Tom Christie:
Rob: Alvy, you’re a total paranoid.
Alvy: Wh — How am I paran—? Well, I pick up on those kinds o’ things. You know I was having lunch with some guys from NBC, so I said … uh, “Did you eat yet or what?” and Tom Christie said, “No, didchoo?” Not, did you, didchoo eat? Jew? No, not did you eat, but jew eat? Jew. You get it? Jew eat?
Puns are often the despair of translators because they depend on sounds of the source language that the target language does not share. The standard Spanish subtitles for Annie Hall decline to imitate the pun, instead having Christie say: “No. ¿Y tú, judío?” (“No. Did you, Jew?”) This represents Christie’s response as a blunt insult, which is not exactly how it appears in Allen’s screenplay; the whole point of the pun is that, while it could sound like a blunt insult to Alvy, he is potentially overreacting. Venuti contrasts the rather flat-footed rendering of the pun with the one offered by José Luis Guarner in his own translation of the screenplay — “Tom Christie me respondío: ‘Sí, judías.’”— and explains why this is an ingenious solution:
The word ‘judías’ perfectly fits the context, the lunchtime conversation between Alvy and Tom Christie, because it signifies a food in peninsular Spanish, “green beans.” Yet because the word can also signify “Jewish women,” a meaning activated by the topic of Alvy’s conversation with Rob, it works as a pun and conveys the racist innuendo that Alvy detects.
Of course, Guarner’s translation reinterprets the meaning of Christie’s statement quite dramatically (“no,” for example, becomes “yes”). And, to be sure, “judías” does not replicate the meaning or effect of “didchoo/Jew.” It can’t do this, since “judías” is actually both a word for food and a potential slur, while “did you” and “Jew” are different words that happen to sound the same, given certain pronunciations and sensitivities. And Venuti points out that “judías” will mean something different to, say, speakers of Spanish on the peninsula versus those elsewhere. All of this is important. Nevertheless, Guarner’s translation, even as it deviates from the literal meaning of the source text, brilliantly conveys Allen’s intention that there be a play on words in this scene. He is not hemmed in by a straitjacket of fidelity or semantic equivalence to his source. Moreover, as Venuti explains, Guarner’s translation does more than merely convey Allen’s intention: “it exposes the fact that the English treats anti-Semitism as a form of paranoia that can become the basis of jokes, even as the English comes back to worry the equivalence of the Spanish version and to point up the pressure to represent an instance of persecution — perhaps not surprising in a translation published within a decade after the end of Franco’s dictatorship.”
Surely Venuti is right that Guarner’s translation throws critical light on how two different cultures and two different historical junctures might treat anti-Semitism. And surely he is right, more broadly, when he claims that translations “can be seen as setting up a critical dialectic with the source material whereby they submit one another to a probing critique — although only when approached with the hermeneutics of suspicion.” Yet none of this requires that we adopt a dubious categorical distinction between models of translation or an even more dubious claim about translations taking the place of their sources. To have a hermeneutics of suspicion, you first need a hermeneutics. This can only proceed on an understanding of a source and its language, and on maintaining the link between a source and its translation. Everything wise that Venuti says about the translation of this scene from Annie Hall depends on our ignoring the theories to which his book subscribes. The pun does not vary; the context and the approach to translating it do. Rather than choosing between the instrumental model and the hermeneutic model, every translator, like Guarner, interprets and represents a source. It makes little sense to think of translation as a matter of rejecting one model for the other, or as a matter of replacing a source text with a substitute.
Little sense, that is, unless what we are interested in is really something beyond translation as a practice. The confusions of Contra Instrumentalism — with regard to distinctions of degree and distinctions of kind, and around the relationship of source texts to translations — are related to the book’s enormous ambitions, ambitions that would have us treat translation and translators as if they possessed more autonomy than they do. In reality, the autonomy of translators, like that of interpreters, is circumscribed in a fundamentally different way than it is for those who are not translating or interpreting. Our work will always derive from, and be responsible to, someone else’s. There’s no shame in recognizing this, however much we might want things to be otherwise. Rather than overselling our autonomy, we should follow the example of Venuti’s practical criticism and continue to show how attention to translation is crucial to the larger projects of producing art and knowledge.
The author is grateful to the editors of Provocations and to Joel Calahan, for comments on an earlier draft of this essay, and to Lawrence Venuti, for the invitation to engage with his book.
 I include here for comparison an excerpt from the Italian version of the screenplay, by Benigni and Vincenzo Cerami:
Guido si affianca al corporale che immediatamente si rivolge ai nuovi prigionieri con voce stentorea.
Caporale (in tedesco): Fate tutti attenzione! Parlo una volta sola!
Guido: (con lo stesso tono militare): Comincia il gioco… che c’è c’è, chi non c’è non c’è!
Caporale (in tedesco): Siete stati portati in questo campo per una sola ragione.
Guido (traduce): Si vince a mille punti… il primo classificato vince un carro armato vero…
Caporale (in tedesco): Dovete lavorare.
Guido: Beato lui!
I prigionieri faticano a capire di che cose stia parlando il caporale. Solo Giosuè capisce. È immobile con gli occhi spalancati.
Il caporale indica lo spiazzo fuori.
Caporale (in tedesco): Il sabotaggio sul lavoro sarà punito con la morte, qui davanti sul cortile mediante fucilazione alla schiena.
(E indica la propria schiena.)
Guido: Ogni giorno vi daremo la classifica generale… là da quell’altoparlante che sta fuori! All’ultimo classificato attaccheremo un cartello con su scritto “asino”… qui nella schiena!
(E imita il gesto del corporale.)
Caporale (in tedesco): Avete il privilegio di lavorare per la grande Germania, per la costruzione di un grande impero!
Guido: Noi facciamo la parte di quelli cattivi cattivi che urlano. Quello che hanno paura perdono punti!
Roberto Benigni and Vincenzo Cerami, Life is Beautiful, Trans. Lisa Taruschio (New York: Hyperion, 1998) (130–131). Taruschio’s translation maintains semantic equivalence with the source while also using functionally equivalent idiomatic expressions: “Lucky man” for “Beato lui,” for example, and “mean guys” for “cattivi cattivi.” An aspect of the source that does not come through in her translation occurs in the stage directions, where there is a play on “stentorea“: unlike the legendary herald Stentor, whose powerful voice ensured it could be heard, the German corporal’s message is obscured by the language barrier and by Guido’s creative “translation.”
 Sander Gilman, “Is Life Beautiful? Can the Shoah be Funny? Some Thoughts on Recent and Older Films,” Critical Inquiry 26, no. 2 (2000): 279–308 (292).
 George Steiner, After Babel: Aspects of Language and Translation (Oxford: Oxford University Press, 1975) , 251.
 Steiner, After Babel, 436.
 Lawrence Venuti, Contra Instrumentalism: A Translation Polemic (Lincoln: University of Nebraska Press, 2019), 39. Ironically, Venuti presents these technical philosophical terms in English with no translations. This is particularly significant (and problematic) in the case of “assemblage,” which, as the standard English translation of the French term agencement, has been criticized as narrowing Deleuze and Guattari’s intended meaning and as fostering a confusion with assemblage, which Deleuze and Guattari also use, albeit in a non-philosophical way. See John Phillips, “Agencement/Assemblage,” Theory, Culture, and Society 23, no. 2–3 (2006): 108-109.
 Venuti, Contra Instrumentalism, 39.
 Venuti, Contra Instrumentalism, 39.
 Venuti, Contra Instrumentalism, 1.
 Venuti, Contra Instrumentalism, 1.
 Venuti, The Translator’s Invisibility: A History of Translation (London: Routledge, 1995).
 Gilles Deleuze and Félix Guattari, Anti-Oedipus: Capitalism and Schizophrenia, Trans. Robert Hurley, Mark Seem, and Helen R. Lane (New York: Penguin, 2009), 53. The original reads: “Nous ne prétendons nullement reprendre une tentative comme celle de Malinowski, montrant que les figures varient d’après la forme sociale considérée. Nous croyons même à ce qu’on nous dit lorsqu’on présente Œdipe comme une sorte d’invariant. Mais la question est tout à fait ailleurs : y a-t-il adéquation entre les productions de l’inconscient et cet invariant (entre les machines désirantes et la structure œdipienne) ?” Gilles Deleuze and Félix Guattari, Anti- Œdipe: Capitalisme & Schizophrénie, Nouvelle Édition Augmentée (Paris: Éditions de Minuit, 1973), 61.
 Venuti, Contra Instrumentalism, 2.
 Deleuze and Guattari, Anti-Oedipus, 53.
 Venuti, Contra Instrumentalism, 3.
 Venuti, Contra Instrumentalism, 3.
 Venuti, Contra Instrumentalism, 2.
 It would be possible for Venuti to contest this sharp division between distinctions of degree and distinctions of kind by appealing to Deleuze’s ontology, in which the sharpness of this distinction is lost. However, this appeal would concomitantly dull the edges of “radical” and thus present another problem for the argument Venuti advances.
 Venuti, Contra Instrumentalism, 3.
 Ludwig Wittgenstein, Philosophical Investigations, rev. 4th ed., trans. G.E.M. Anscombe, P.M.S.
Hacker, and Joachim Schulte (Oxford: Blackwell, 2009), 53e.
 For representation as denoting, expression, and/or exemplification, rather than “imitation,” see Nelson Goodman, Languages of Art (Indianapolis: Hackett, 1976), especially chapters 1 & 2.
 Ezra Pound, Literary Essays (New York: New Directions, 1968), 232.
 Venuti, Contra Instrumentalism, 93.
 Venuti, Contra Instrumentalism, 93.
 Venuti, Contra Instrumentalism, 17.
 Venuti, Contra Instrumentalism, 18.
 Venuti, Contra Instrumentalism, 19.
 Venuti, Contra Instrumentalism, 146.
 Venuti, Contra Instrumentalism, 147.
 Venuti, Contra Instrumentalism, 149.
 Venuti, Contra Instrumentalism, 149.