Anthony Pym


_Video material

Doubts about deconstruction as a general theory of translation

Anthony Pym

Published in TradTerm (São Paulo) 2 (1995), 11-18. Passages in red have been added for clarification in view of subsequent responses.

Abstract: Comparison of four versions of a sentence from Derrida poses the question of how deconstructionist philosophy should be related to translation theory. Doubts are raised with respect to the general pertinence of deconstruction, the possibility of extending its insight beyond source-text analysis, and the reasons for a certain residual inferiorization of translation. It is suggested, in the spirit of quiet dissent, that translation theory should not be unduly upset by the fact that source texts are semantically unstable points of departure.

Andrew Benjamin begins his Translation and the Nature of Philosophy by citing and translating Derrida:

"With the problem of translation we are dealing with nothing less than the problem of the passage to philosophy." (1989:1)

To talk about translation would thus be to talk about philosophy, and perhaps vice versa. Little of this obviously very important discussion would be held up by petty suspicion of the way Benjamin has translated this claim from French, nor by the significantly different version on the back cover of the book in question:

"With the problem of philosophy [sic.] we are dealing with nothing less than the passage to philosophy."

We've all been victims to similar slips. To make much of them would be decidedly below the belt. Yet some variants, as deconstructionists know, mean more than they mean to. The secret of this one might be found through Barbara Johnson's version of the same sentence from Derrida's La dissémination:

"With this problem of translation we will thus be dealing with nothing less than the problem of the very passage into philosophy." (1981:72)

The differences are surely trivial. There is no overwhelming reason why two or three disseminations of Derrida should come out even nearly the same. Benjamin is translating an authoritative starting point for his "new theory of words;" he needs as much of an axiom as non-axiomatic thought can provide; he has little time for deictics that might mislead away from the point, be it backwards ("this problem of translation," says Johnson) or indeed forwards ("we will thus be dealing with . . ."). The only real site to be cited should be Derrida's apparent imprimatur. And if the words concerned are to function with any authority they must be self-sufficient in the present tense, present as a point of departure for the new translator-philosopher. Benjamin's situation thus requires more than overtly interpretative reported speech ("Derrida suggested the problem of translation was . . ."); his starting point is well served by translation in a stronger sense, where the first person belongs to Derrida himself. And this translation would have worked, no doubt, had its status not been tainted by the unfortunate back cover (which arouses the question of what Derrida really said) and, less unfortunately, by Johnson's previously available version, itself re-cited and prominently disseminated in the volume edited by Graham (1985:146). Only in the absence of rivals can a translation really function as a solid point of departure. Only in the absence of variants is it really received as translation rather than economically reported speech. We shall return to this in a moment.

Johnson's position is different. Translating a book, or more exactly an essay on certain problems of translation, she has every interest in pointing both backwards and forwards to the connections in a typically arabesque argument. She helps the reader orient the point. The "this" of her "this problem of translation" refers to a rather specific difficulty:

." . . situated less in the passage from one language to another, from one philosophical language to another, than already, as we shall see, in the tradition [sic.] between Greek and Greek; a violent difficulty in the transference of a non-philosopheme into a philosopheme. With this problem of translation we will thus be dealing with nothing less than the problem of the very passage into philosophy." (1985:146)

The problem referred to, now a piece of deconstructionist folklore, is Plato's use of the word pharmakon as at once a common term (non-philosopheme) and a technical philosophical term (philosopheme), as at once a polysemous part of a language (both "cure" and "poison" in Greek) and as a unitary proper name (for a concept in Plato). The example contains most of what Derrida's translators have to say about translation. "This problem of translation" is thus rather specifically the traditional transfer leading to philosophical terms, no matter what language we are working in or from. One might say it was really the problem of philosophy tout court, in which case Derrida would of course be saying, "With the problem of philosophy we are dealing with nothing less than the passage to philosophy." So perhaps Benjamin's back cover is not entirely antipodean.

Justification for the back cover can be found in a further translation of the same passage, this time into Rosemary Arrojo's Brazilian Portuguese:

"Ao lidarmos com o problema da tradução, estamos lidando com nada menos do que o problema da passagem para a filosofia." (1993:75)

The version possibly owes much to Benjamin's English. But now we find a new ambiguity. The Portuguese "para" would normally be read as "to," rendering the idea of a "passage to philosophy." However, with a pause after the term "passagem," it could also be read as "for," suggesting that the real problem here could be "for philosophy." The latter reading, although willfully perverse [and despite the fact that this is how my Brazilian students first translated it into English, before being questioned more closely], seems to find support in Arrojo's gloss (if I may translate):

"If the passage of a signified from one language to another, from one context, time and place to another context, time and place, is irredeemably marked by difference and displacement, if that which is supposed to be the 'same' is revealed to be 'different' and 'other,' in an implacable game, how can philosophy protect its inaugural interest in the possibility of a 'passage' for a universal truth, anterior and exterior to any language?" (1993:75).

This makes sense. But it does not make the same sense as Benjamin's and Johnson's reference to a "passage to philosophy." The English-language versions concern translation within Greek. Arrojo's problem here is the translation of philosophy from one language to another (although the in-Greek problem is translated, explicitly from English, on page 207 of her book just cited). Within this more conventional frame, the gloss asks a question that is fairly easily answered. Philosophy could protect itself from translation by not being translated, by aspiring to the formalized language of music, logic or mathematics. Alternatively, it might make its readers learn languages. Derrida went back to the Greek to recover a problem lost (or solved?) in translation. And philosophers might at this stage be tempted to go back to Derrida's French, to find whatever the above translations might have displaced. Deconstruction at once invites such a move (to find the difference) and condemns it as logocentric (the initial search would be for a stable meaning that is not to be found). One can only deconstruct an attempted construction.

Arrojo's version of Derrida's sentence eclipses the passage from Greek to Greek. Her translation seems peculiarly unworried by any return to original languages or anterior contexts. The resulting approach can even deconstruct Laplanche's translation of Freud without any visible return to the French or German texts (1993:35-50). Such analysis admirably escapes the aspiration attributed to philosophy. Translation is not logocentric philosophy, and all philosophy is apparently logocentric (but what of twentieth-century philosophies of science based on falsifiability instead of truth?). This particular translation moreover performs the displacement it describes. It could be of considerable value. After all, such a passage of philosophy could only be a problem for a discourse that wanted to remain the same. It could only be a problem for philosophy as defined.

If translation is not philosophy, there is no reason why translation theory should be embedded in a philosophy. No search for stable transfer need be involved in any categorization of the various versions of Derrida. We could say, for instance, that Benjamin's back cover looks like a mistake; his in-text version could be a contextually calculated use of error; Johnson's work might be something like translation with a reader in mind; Arrojo's reading simplifies the problem for the sake of pedagogical or evangelistic interests. We would thus be talking about different kinds or levels of translation, without presupposing any transfer of stable meaning.

Yet deconstructionists would no doubt find these charitable categories unacceptably normative. There are no lines between classes, and each class norm would falsely presuppose a correct version from which the actual results diverge to various degrees. Deconstruction correctly wants none of this. As far as possible, all outcomes come out from an anterior plurality. So one must proceed differently with the differences involved.

Can Johnson's "this problem of translation" also be Benjamin's and Arrojo's "the problem of translation"? The passage to philosophy (from non-philosopheme to philosopheme), just like the passage of philosophy (from one context to another), is certainly a problem of philosophy (could it be anyone else's?) and thus of translating philosophy. But does anything guarantee it is a problem of all translation, let alone the problem of everything translators do. Is it the only problem of translation, nor indeed a problem only of translation? The only generality firmly attributed to this problem is of a kind accrued by negation, through the fact that other approaches to translation appear to ignore it. The argument is not entirely perverse: If everyone else fails to perceive a truly problematic problem (non-philosopheme to philosopheme, stable meaning across contexts), then the problem can be as general as the perception that fails to see it. Let's see how this works.

Benjamin's use of this argument involves repeated reference to something called "the conventional understanding of translation," "translation in its most naïve understanding" (1989:60) and so on, implying a mode of perception that is at once general and yet unable to perceive a problem that is just as general. The discursive strategy shares the same circuit as the translated phrase. Derrida is similarly wont to disdain "le concept courant de la traduction" (1985:221); Johnson is ironic about translation as it has "always been" (1985:145); Arrojo attacks the "common sense and theories of language" that presuppose fidelity to originals (1993:28). These references all project a benighted notion of translation as the transfer of stable meaning. Deconstruction can thus enter to dispel darkness and confusion. What you thought was transfer is revealed to be transformation; the generally accepted notion was wrong, so the previously unseen problem must rightly be general. This is similar to what students are told as they enter some German translation institutes: Whatever you think you know about translation, you're wrong. But here, in deconstruction, the new generality is of rather larger dimensions. It must at least be postmodern, if not post-Enlightenment, post-Christian, post-western-metaphysics, post-logos or even posted, belatedly, from Paris.

Such a scenario requires quite massive ignorance of the the many historical alternatives to theories of translation as a transfer of stable meaning. Deconstructors are correct to react against simplistic theory. But they should first locate their opposition with some precision. Meaning transfer is, to be sure, a common basis for twentieth-century linguistic theories of translation, especially those suited for mechanizing translators. The notion has further been assisted by the development of service ethics and a broad feminization of the translator's profession. However, as we have seen in Benjamin's need to translate a point of departure, meaning transfer can be something more than a theorist's theory. It is also an assumption made in any reading of a translation as a translation. Rival translations are less translational precisely to the extent that they challenge assumptions of stable meaning. If, for example, we do not read French and have no access to Derrida's text, four different versions of the same phrase can only raise doubts because some meaning transfer is presupposed. If this were not the case, no questions would be asked and the texts would be consumed indifferently as translations or non-translations.

Meaning transfer is thus an assumption - certainly a social illusion - operative in the use of translations as translations. Yet it is not ubiquitous. Inasmuch as there are users for whom the source text is unavailable, this assumption of meaning transfer is specifically external to actual translation processes. No translator or translation critic need believe that translation is the transfer of stable meaning. Indeed, inasmuch as there is a plurality of translators for whom source-text meanings differ, internal knowledge of translation is quite likely to accept the initial plurality of translation processes. Users have to believe in meaning transfer, but most translating translators are quite likely to accept deconstructionist arguments about the instability of their sources.

Just as deconstruction requires a search for original meaning in order to reveal differences, so translation requires the external reader's belief in meaning transfer in order to produce internal differences. Yet the deconstructive difference is something that happens in the move to and of philosophy. The translative difference concerns moves between internal and external social positions.

The deconstructionist insult comes from confusing socially operative external illusions with professionally retained internal knowledge. "This" problem can become "the" problem only from the perspective of users who have no access to the specificity of "this" except through translation. Internal knowledge is arguably aware of its initial plurality. So what can deconstruction say that most practising translators don't know already? And what can it say to actual users of translations?

Complexity, if nothing else, must be a feature of whatever could be said. If deconstructionist argument were interested in the four versions of Derrida's sentence, its prime reaction would of course be to go back to what Derrida actually wrote, if not further back to some antecedent in German or Greek. The first move towards complexity would perhaps be to leaf through at least the 1972 French text of La dissémination, reaching page 80, recapturing some of the master's argument and perhaps concluding that neither Benjamin, Johnson nor Arrojo have fully recuperated the play of Derrida's significance. Needless to say, anyone engaged in this process has no social need of the translations. They might appreciate the various versions as extensions of Derrida's game, but such appreciation is the privilege of an internal position. Those who need the translations, in their admittedly relative externality to differential epiphanies, could only go back by learning French, reading Derrida "in the text" and occupying internal positions. And as such enlightened readers, now transformed from translation-users into amateur philosophers, as they ask what Derrida really said - but for whom should it matter? -, the complex answers they receive might function as caveats but are more likely to incur unprofessional frustration, or frustration with the translator's (non)profession. An essentially internal discussion about the plurality of translation, held between translators or those who presume to know about translation, cannot be expected to enthuse someone who necessarily assumes and uses translation as meaning transfer.

There is some irony in the way that the critique of origins tends to invest all its efforts on the level of origins, to the detriment of efficient, formal or final purposes (bearing in mind that Aristotle's categories displaced the pre-Socratic fixation on initial causes). A critique of origins is inevitably locked into a backward vision, to the detriment of present action or future agreements. In its psychoanalytic metaphors, this critique focuses on the imaginary status of initial causes but forgets that the analyst's final cause is to help cure someone. Deconstruction might perhaps be able to say something about how a translator should accept semantic plurality - hopefully beyond the luxuriously expensive solution of perpetual retranslation -, or how a user should assess a translation in terms of this plurality. But as soon as the specific problem of anterior origins becomes "the" problem of translation, deconstruction reduces translation to a form of source-text analysis. In fact, it turns translation into what could only be an inferior form of the kind of readings undertaken by deconstruction itself. Translators would be people who should write literary or philosophical critiques but are unable to, lacking textual space, paid time and perhaps sophistication. They can only introduce real philosophers to the passage to real philosophy.

This could explain the peculiar way deconstructionist texts tend to inferiorize translation, even when they have no real interest in doing so, even when they proclaim they are doing precisely the opposite. In his early work Derrida made a point of saving écriture from the status of mere translation, "a technique in the service of language, a porte-parole" (1967: 17-18). He even defined translation as "a wholly linguistic movement [i.e. not working from thought] ensuring the transportation of a signified from one language to another" (1972b: 226). Kinder and more interesting things were later said (possibly for reasons suggested below). But actual translations are nevertheless mostly seen in a rather traditional way, often saying less than their antecedents. Although not as explicit as in Heidegger, considerable inferiorization remains in Derrida. It is a kind of devaluation inherent in philosophy as a discourse on great philosophical texts, at worst confused with a history of all texts, thus reducing translation to work on the same great texts. When "this problem" becomes "the passage to philosophy," the resulting theory follows Schleiermacher's "Methods of Translating" (1813) in eliminating from consideration all the lesser tasks of commerce, negotiation and interpretation, mere facts of everyday life. Since the only kind of translation worth talking about becomes (written) work on great texts, no translation worth talking about is considered better than the plurality of its original, by definition accorded the textuality of greatness. Philosophers have no real time for the rubbish that most of us have to improve when we translate.

These then are three related elements of what might be called a certain deconstructionist approach to translation: opposition to a generalized external conception of translation, restriction to a problematic of origins, and a residual inferiorization of translation.

This is not to say that the deconstructionist problem has nothing to do with translation. Most translators would enjoy being called sites of difference, plurality, conflict, rupture, productivity, even of translativity anterior to the work paid for as translation. Anything that can be translated as stable meaning transfer is best left to machines anyway. But having agreed with this, are we any better off? Do we know how to translate, when and where to translate, or how to get paid for translating and retranslating? Can we make our internal knowledge compatible with external criteria that require us to limit the range of interpretations we present?

The problem with deconstructionist propositions, like those based on the obverse dominance of target-side or final causes, is that they are decidedly unhelpful once agreed to. One eventually has to ask "So what?" and then get on with solving concrete problems. But few concrete problems are addressed by the strangely binary way in which Benjamin indicates two possible directions then returns to the labyrinth:

"If it is no longer possible to provide the conditions of existence for the possibility of translation in terms stemming from rational recovery - where what is recovered and re-expressed is the original content of the original text - then the emphasis must shift to the text itself and hence to a concern with language." (1989:86)

Not at all. There are many other places requiring attention, with or without the actual possibility of meaning transfer. Translation theorists could ask, for example, how it is possible for anyone to believe in rational recovery, why translation exists even when this condition is known to be illusory, how translators actually proceed despite source-text plurality, for whom they reduce or convey that plurality, why plurality could or should be reduced or conveyed or augmented, at what social cost, when, where and by whom. None of these questions can be addressed by focusing on the text itself or indulging in nebulous mysteries of language. All of them require awareness that translation actually does something in and of itself.

Original textuality is just an initial problem for translators. Since theirs is an action that is to become a starting point for future actions, just as Benjamin's translation from Derrida becomes his own starting point, translators must weigh up the costs of conveying relative complexity. They have to make choices between available alternatives, many of which involve potentially beneficial reductions of plurality. Such choices require ethical guidelines. And yet the question of ethics, which is where translation theory needs real help from philosophers, is precisely the point obscured by fixations on origins.

[Cf. at this point Arrojo's gloss on the last sentence (1997: 5-24): "Pym suggests that the task of devising an appropriate ethical code for translators should be conferred upon philosophers, provided that such philosophers do not get mixed up with Derridean deconstruction, that is, provided that such philosophers rely on the possibility of stable 'originals', stable 'laws', and, of course, the possibility of imposing them worldwide and for all time." Where the hell did she get that from? Will someone please teach this woman to read? Now back to the action...]

Witness Benjamin's sad deconstruction of Davidson's very elegant recommendation that our attributions be charitably rational. Davidson could mean that contradictory back-cover variants should be overlooked as mistakes; in-text simplifications should be understood in terms of their argumentative function; orientative deictics should be appreciated as space for readers; simplification should be regarded as a pedagogical virtue. In each case, what is attributed to the source is rational in terms of the receptive situation, regardless of whatever plurality or irrationality might have actually been at the source (the truly external user of the translation, who ultimately makes the attribution, does not know the source anyway). But Benjamin, whose translative endeavors require at least this charity, obscures Davidson's originality by reducing attribution to an unrequired Kantian universalism, converting a present-tense ethical recommendation into an historical belief in origins. He thus eternally returns to the only problem that deconstruction was equipped to deal with in the first place. His deconstruction heads back to the labyrinth because it cannot see anywhere else to go.

[Interestingly, Arrojo's citations charitably integrate Davidson's charity (1993: 64-64) and concordantly describe deconstruction as a "new humanism" (92). The Brazilian Derrida might eventually find somewhere else to go.]

Translation theory currently faces many important challenges: to envisage intercultural relationships that presuppose neither sovereignty nor hegemony; to construct a workable language regime for multicultural empires; to define the authority of intermediaries and their ethical capacity to intervene; to work productively with machines; to develop intercultural communities and transnational training programmes, hopefully without paternalistic impositions southward and eastward; and most generally, to achieve understandings, not in the sense of returning to any naturally established rational universality but certainly in the sense of directing attributions in and for the present and future, as contracts and treaties, potentially indifferent to different origins. For example.

In a world where everything is to be constructed, deconstructionist theory can and should raise passing doubts on the way to concrete action. But translation theory has a lot of other work to do, awaiting the philosophers' return from Derridean islands.


Check Derrida's text (page 80 of the 1972 Minuit edition). It doesn't really matter which translation was closest to the master. There are interesting things to be found in the relations between the translators themselves.

On the page in question Derrida certainly pays brief homage to the difficulty of translation. But his chief source of delight here, and again 21 pages further on in the text, is his ability to correct the authority of the previous translators, whose edition of Plato "fait autorité." The previous translators effaced the duplicity of "pharmakon;" they were not good philosophers. Or rather, Derrida the philosopher can show himself to be a better translator. The passage to philosophy is a matter for new French philosophers, not for authoritative French translators of past generations. In this context some attention should also be paid to the place and date of the text's first publication: Tel Quel, 1968 (our passage is on page 9 of the first installment). The questioning of authoritative translators was also a questioning of French authority. Derrida's text was of its age. But it may not be immediately applicable to subsequent conjunctures.

Translated into American English by Barbara Johnson in 1981 and commented upon by her in 1985, Derrida's "Pharmacy" became something more than a contestation of official translators. In the United States, the established New Critical emphasis on reading made Derrida a theorist of reading, and thus an ally of translators as writers of critical readings. The American Derrida was not the Derrida of 1968: Johnson stresses the difficulty of translation, underplaying the irony of the French Plato that "fait autorité." Further, I suggest, Derrida responded to this new market location. He wrote directly on translation and directly for his own (American) translators.

When Benjamin cited Derrida in Britain in 1989, he did not do so as an acolyte unable to theorize without the master (cf. Benjamin 1976). He was instead indicating an established academic location with a certain contestational value in the English-speaking world. Indeed, he might have been trying to tap an established deconstructionist market, very belated in its arrival in Britain via a long American detour. Hence the prominence of the translated citation.

Translated from American English into Brazilian Portuguese by Arrojo in the early 1990s, Derrida [I mean the Derrida of the sentence under analysis] is something else again. He has become a "guru for a whole generation of scholars [...] at the Sorbonne, the Ecole Normale Supérieure, Yale, Harvard, the University of London, and most of the linguistic and philosophical centers of universities in the developed world [do mundo desenvolvido]" (Arrojo 1992: 10). Really? To read and translate Derrida in Brazil is a sign of development, a sign of how translation should be in the developed world, a sign to be pedagogically simplified and imitated if one is to catch up.

As Baudrillard said of marxism (1973: 38), Europe exports its ideas when they no longer work at home. When post-1989 Europe is self-deconstructing without any visible help from philosophers, other parts of the world can accept deconstruction as a postmodernist theory of translative reading, or even as a very modernist sign of development, on a par with buying Europe's old trains and airplanes. But post-1989 is not 1968.


* A version of the first sections of this paper was published in the proceedings of the 1993 FIT conference held at Brighton. The final version was written with the help of a research fellowship from the Alexander von Humboldt Foundation for work carried out at the Sonderforschungs-bereich 309 "Die literarische Übersetzung" at Georg-Universität Göttingen, Germany.

So does Anthony Pym support the idea of translation as meaning transfer? Not at all. The book Translation and Text Transfer(1992) builds from the notion of materially moving signifiers, not signifieds, and borrows its materiality from the initial Derridean gramme. That book also has a section called "Translation theory should not lecture translators", by the way.


Arrojo, Rosemary. 1992. "Apresentação." In Arrojo (ed.) O Signo Desconstruído. Implicacões para a tradução, a leitura e o ensino. Campinas São Paulo: Pontes. 9-12.

Arrojo, Rosemary. 1993. Tradução, Desconstrução e Psicanálise. Rio de Janeiro: Imago.

Arrojo, Rosemary. 1997. "Asymmetrical Relations of Power and the Ethics of Translation." TextconText 11/1. 5-24.

Baudrillard, Jean. 1973. Le Miroir de la production ou l'illusion critique du matérialisme historique. Paris: Castermann.

Benjamin, Andrew. 1976. "A Theory of Reading with Respect to Freud and Lacan." Working Papers in Sex, Science and Culture. Darlington (Australia), 1/1.

Benjamin, Andrew. 1989. Translation and the Nature of Philosophy. A New Theory of Words. London and New York: Routledge.

Derrida, Jacques. 1967. De la grammatologie. Paris: Minuit.

Derrida, Jacques. 1968. "La pharmacie de Platon." Tel Quel 32/33: 3-48, 18-59.

Derrida, Jacques. 1972a. La dissémination (includes "La pharmacie de Platon"). Paris: Minuit.

Derrida, Jacques. 1972b. Marges de la philosophie. Paris: Minuit.

Derrida, Jacques. 1981. Dissemination. Trans. Barbara Johnson. Chicago: University of Chicago Press.

Derrida, Jacques. 1985. "Des Tours de Babel." In Graham (ed.). 165-207.

Graham, Joseph F. (ed.). 1985. Difference in Translation. Ithica and London: Cornell University Press.

Johnson, Barbara. 1985. "Taking Fidelity Philosophically." In Graham (ed.), 142-148.

Schleiermacher, Friedrich. 1813. "Über die verschiedenen Methoden des Übersetzens." Sämtliche Werke, Dritte Abteilung (zur Philosophie). Vol. 2. Berlin: Reimer, 1838. 207-245.

Last update: 8 June 1999



© Anthony Pym 2014
URV. Av. Catalunya, 35
45002 Tarragona, Spain
Fax: + 34 977 299 488