Natural and directional equivalence

in theories of translation

Anthony Pym
Universitat Rovira i Virgili, Tarragona, Spain*

Equivalence was a key word in the linguistics-based translation theories of the

1960s and 1970s, although its basic mode of thought may be traced back to
Cicero and later to the Renaissance theories that began to presuppose languages
of equal status. Close inspection reveals that some theories assume pre-existing
equivalents and are thus concerned with a search for natural equivalence. Other
theories allow that translators actively create equivalents, and are thus concerned
with directional equivalence. The first kind of equivalence is concerned with
what languages ideally do prior to translation; the other deals with what they
can do. These two approaches are often intertwined, giving rise to many mis-
understandings and unfair criticisms of the underlying concept. The historical
undoing of the equivalence paradigm came when the directional use of the term
allowed that equivalence need be no more a belief or expectation at the moment
of reception, which need not be substantiated on the level of linguistic forms. At
the same time, source texts became less stable and languages have been returning
to more visibly hierarchical relations, further undermining the concept. Contem-
porary localization projects may nevertheless fruitfully be interrogated from the
perspective of natural and directional equivalence, since the presumptions are
being used by contemporary technology precisely at the moment when the terms
themselves have been dropped from critical and exploratory metalanguage.

Keywords: directional equivalence, lexical semantics, localization, natural

equivalence, relevance theory, translation strategies, types of equivalence

Parable
At one stage in the criminal trial of O.J. Simpson, a photo was shown of backyard at night,
with the killers footsteps visible in the moonlit dew. A Charlie-Chan detective then scruti-
nized the photograph. Over there, more dimly in the dew, he saw another set of footsteps.
Two paths, not one. So which footsteps were the killers? And for that matter, who took the
photo, and how did they get there?

Target 19:2 (2007), 271294.

issn 09241884 / e-issn 15699986  John Benjamins Publishing Company
272 Anthony Pym

1. Introduction to a historical location

The term equivalence, in various European languages, became a feature of West-

ern translation theories in the second half of the twentieth century. Its heyday
was in the 1960s and 1970s, particularly within the frame of structuralist lin-
guistics. The term roughly assumes that a source text and a translation can share
the same value (equi-valence) on some level, and that this assumed sameness is
what distinguishes translations from all other kinds of texts. Within that para-
digm, to talk about translations was to talk about different kinds of equivalence.
That might still be the case. Equivalence, however, has since been surpassed by a
range of alternative paradigms. It became no more than a special-case scenario
for German-language Skopostheorie; it was a banal constant for Tourys Descrip-
tive Translation Studies; it has long remained at loggerheads with indeterminism,
and its implicit essentialism has become an easy pop-shot for deconstructionist
approaches. As a result, the equivalence paradigm has come to be regarded as
nave or limited in scope. Mary Snell-Hornby, for example, jettisoned equivalence
as presenting an illusion of symmetry between languages which hardly exists
beyond the level of vague approximations and which distorts the basic problems
of translation (1988: 22).
Here we take the unpopular view that the equivalence paradigm was and re-
mains far richer than the facile dismissals would suggest. Its metalanguage mer-
its a serious place alongside and indeed within the more recent alternative para-
digms. We shall attempt to show that the theorizing of equivalence has in fact
involved two competing conceptualizations: natural as opposed to directional
equivalence. Two paths, not one. The intertwining duality of these notions allows
for considerable subtlety in some past and present theories, as well as pervasive
confusion not only in many of the theories themselves but also in the many argu-
ments against equivalence.
At the same time, we readily recognize that equivalence denominates a pro-
foundly historical paradigm. Notions of equal value presuppose that different
languages do or can express the same values. Such equal potential might have been
possible between classical Greek and Latin, allowing Cicero, resting on Horace, to
conceptualize the one text as being translated in two different ways (ut interpres
vs. ut orator), which necessarily assumes that there is some value that remains con-
stant. That was a fundamental conceptualization of equivalence, albeit without the
term. Nevertheless, that particular belief in equal value across languages was rarely
fronted in European theorizing prior to the Renaissance. The medieval hierarchy
of languages usually meant that translation was conceptualized as a way of en-
riching the target language with the values of a superior source language (most
translations went downwards in the hierarchy, from Hebrew or Greek to Latin, or
 Natural and directional equivalence in theories of translation 273

from Latin to the vernaculars). For as long as the value hierarchy existed, claims
to equivalence (without the term) played little role in thought on translation. For
roughly parallel historical reasons, the conceptual geometry of equivalence was
difficult to maintain prior to the age of the printing press, since printing rein-
forced notions of a fixed source text to which a translation could be equivalent.
Before that fixing, textuality tended to involve constant incremental changes in
the process of copying, such that translation was often just further extension of
that process. Printing and the rise of the vernaculars facilitated the conceptualiza-
tion of equivalence, albeit still without the term (its place was often marked by
talk of fidelity to meaning, intention or function). In accordance with this same
logic, the relative demise of equivalence as a concept would correspond to the elec-
tronic technologies by which contemporary texts are constantly evolving, primar-
ily through text re-use (think of websites, software, and product documentation).
Without a fixed textuality, to what should a translation be equivalent? We retain
the answer for the end of our story.

2. Equivalence as a concept

Most discussions of equivalence concern typical misunderstandings. For instance,

Friday the 13th is an unlucky day in English-language cultures, but not in most
other cultures. In Spanish, the unlucky day is Tuesday the 13th. So when we trans-
late the name of that day, we have to know exactly what kind of information is re-
quired. If we are just referring to the calendar, then Friday will do; if we are talking
about bad luck, then a better translation would probably be Tuesday 13th (actu-
ally martes y 13). The world is full of such examples. The color of death is mostly
black in the West, mostly white in the East. A nodding head means agreement in
western European, disagreement in Turkey. That is all boring textbook stuff, but an
understanding of these differences is an essential part of translating.
The concept of equivalence underlies all these cases. Equivalence, we have
seen, says that the translation will have the same value as (some aspect of) the
source text. Sometimes the value is on the level of form (two words translated by
two words); sometimes it is reference (Friday is always the day before Saturday);
sometimes it is function (the function bad luck on 13 corresponds to Friday in
English, to Tuesday in Spanish). That is why Ciceros two alternative ways of trans-
lating can be considered a basic conceptualization of equivalence. Equivalence
does not say exactly which kind of value is supposed to be the same in each case; it
just says that equal value can be achieved on one level or another.
Equivalence is a very simple idea. Unfortunately it becomes quite complex in
its applications. Consider the television game-shows that are popular all over the
274 Anthony Pym

world. English audiences usually know a show called The price is right. In French this
becomes Le juste prix, and in Spanish, El precio justo. Equivalence here is not on the
level of form (four words become three, and the rhyme has been lost), but it might
be operative on the level of reference or function. In German this show became Der
Preis ist heiss, which changes the semantics (it back-translates as The price is hot, as
when children play the game of rising temperatures when one comes closer to an ob-
ject). But the German version retains the rhyme, which might be what counts more
than anything else. It could be getting very warm in its approach to equivalence.
If you start picking up examples like this and you try to say what stays the
same and what has changed, you soon find that one can be equivalent to many
different things. For example, in the game-show Who wants to be a millionaire?
(which seems to retain the structure of that name, more or less, in many language-
versions), the contestants have a series of lifelines in English, jokers in French and
German, and a comodn (wild-card) in Spanish. Those are all very different images
or metaphors, but they do have something in common. Describing that common-
ness can be a difficult operation. More intriguing is the fact that the reference to
millionaire is retained even though different local currencies make the amount
quite different. Given that the show format came from Britain, we should per-
haps translate the pounds into euros or dollars. This might give Who wants to win
$1,867,500? The title has more money but is decidedly less catchy. One suspects
that equivalence was never really a question of exact values.

3. Equivalence vs. langue

In the second half of the twentieth century, translation theorists mostly dealt with
this kind of problem against the background of structuralist linguistics. A strong
line of thought leading from Wilhelm von Humboldt to Edward Sapir and Benja-
min Whorf argued that different languages expressed different views of the world.
This connected with the vision of the Swiss linguist Ferdinand de Saussure, who
in the early years of the twentieth century explained how languages form systems
that are meaningful only in terms of the differences between the terms. The word
sheep, for example, has a value in English because it does not designate a cow
(or any other animal for which there are names in English) and it does not refer
to mutton, which is the meat, not the animal (the difference between names for
animals and names for their meat is fairly systemic in English) (Saussure 1916:
115). In French, on the other hand, the word mouton designates both the animal
and the meat, both sheep and mutton.
Such relations between terms were seen as different structures. Languages
were considered to be sets of such structures (and hence different systems).
 Natural and directional equivalence in theories of translation 275

Structuralism said we should study those relations rather than try to analyze the
things themselves. Do not look at actual sheep, do not ask what we want to do with
those sheep, do not ask about the universal ethics of eating sheep meat. Just look
at the relations, the structures, which are what make language meaningful. One
should therefore conclude, according to structuralist linguistics, that the words
sheep and mouton have very different values. They thus cannot translate each other
with any degree of certainty. In fact, since different languages cut the world up in
very different ways, no words should be completely translatable out of their lan-
guage system. Translation should simply not be possible.
That kind of linguistics is of little help to anyone trying to translate television
game-shows. It is not of any greater help to anyone trying to understand how the
translations are actually carried out. Something must be wrong in the linguistics.
As the French translation theorist Georges Mounin argued in the early 1960s,
If the current theses on lexical, morphological, and syntactic structures are ac-
cepted, one must conclude that translation is impossible. And yet translators exist,
they produce, and their products are found to be useful. (1963: 5)

Either translation did not really exist, or the dominant linguistic theories were
inadequate. That is the point at which the main theories of equivalence developed.
They tried to explain something that the linguistics of the day could not explain or
somehow did not want to explain.
Think for a moment about the kinds of arguments that could be used here.
What should we say, for example, to someone who claims that the whole system
of Spanish culture (not just its language) gives meaning to martes y 13 in a way
that no English system could ever reproduce? Martes y 13 is the stage name, for
example, of a popular pair of television comedians. Or what do we say to Poles
who once argued that, since the milk they bought had to be boiled before it could
be drunk, their name for milk could never be translated by the normal English
term milk? In fact, if the structuralist approach is pushed, we can never be sure of
understanding anything beyond our own linguistic and cultural systems, let alone
translating the little that we do understand. Even more dubiously, structuralist
theories suggested that people within the one linguistic or cultural system did in-
deed understand all the relations at work around them.
Theories of equivalence then got to work. Here are some of the arguments that
were used to address this cluster of problems:
Within linguistic approaches, close attention was paid to what is meant by
meaning. Saussure had actually distinguished between a words value
(which it has in relation to the language system) and its signification (which
it has in actual use). To cite a famous example from chess, the value of the
276 Anthony Pym

knight is the sum of all the moves it is allowed to make, whereas the significa-
tion of an actual knight depends on the position it occupies at any stage of a
particular game. Value would thus depend on the language system (which
Saussure called langue), while signification depends on the actual use of lan-
guage (which Saussure termed parole). For theorists like Coseriu, those terms
could be mapped onto the German distinction between Sinn (stable meaning)
and Bedeutung (momentary signification). If translation could not reproduce
the former, it might still convey the latter. French, for example, has no word
for shallow (as in shallow water), but the signification can be conveyed by
the two words peu profound (not very deep) (cf. Coseriu 1978). The language
structures could be different, but equivalence was still possible.
Some translation theorists then took a closer look at the level of language use
(parole) rather than at the language system (langue). Saussure had actually
claimed that there could be no systematic scientific study of parole, but theo-
rists like the Swiss Werner Koller (1979) were quite prepared to disregard the
warning. If something like equivalence could be demonstrated and analyzed,
then there were systems beyond that of langue.
Others stressed that translation operates not on isolated words but on texts,
and texts have many linguistic layers. The linguist John Catford (1965) point-
ed out that equivalence need not be on all these layers at once, but could be
rank-bound. We might thus strive for equivalence to the phonetics of a text,
to the lexis, to the phrase, to the sentence, to the semantic function, and so on.
Catford saw that most translating operates on one or several of these levels, so
that in the course of a text, equivalence may shift up and down the rank scale
(1965: 76). This was a comprehensive and dynamic theory of equivalence.
A related approach, more within lexical semantics, was to list all the functions
and values associated with a source-text item, and then see how many of them
are found in the target-side equivalent. This kind of componential analysis
might analyze mouton as + animal + meat young meat (agneau), mutton
as + meat young meat (lamb), and sheep as + animal, and then we would
make our translation selections in accordance with the components active in
the particular source text. In the same way, lifeline could be turned into some-
thing like amusing metaphor + way of solving a problem with luck rather
than intelligence + no guarantee of success + need for human external support
+ nautical. We would then find that the translations joker and wild-card re-
produce at least three of the five components, and would thus be equivalent to
no more than that level. There could be no guarantee, however, that different
people would all recognize exactly the same components.
 Natural and directional equivalence in theories of translation 277

All of those ideas were problematic in their own ways. All of them, however, named
or implied a relation of equivalence, and they did so in a way that defended the ex-
istence of translation in the face of structuralist linguistics. Their confrontational
virtue is not to be belittled.

4. Directional vs. natural equivalence

That collection of ideas formed the basis of what might be called the equivalence
paradigm. From the late 1950s, most definitions of translation have consequently
referred to equivalence in one form or another, especially within the field of ap-
plied linguistics. Here are a few of the earlier definitions (italics ours):
Interlingual translation can be defined as the replacement of elements of one lan-
guage, the domain of translation, by equivalent elements of another language, the
range [of translation]. (Oettinger 1960: 110)

Translation may be defined as follows: the replacement of textual material in one

language (Source Language, SL) by equivalent material in another language (Tar-
get Language, TL). (Catford 1965:20)

Translating consists in reproducing in the receptor language the closest natural

equivalent of the source-language message. (Nida and Taber 1969: 12; cf. Nida
1959: 33)

[Translation] leads from a source-language text to a target-language text which is

as close an equivalent as possible and presupposes an understanding of the content
and style of the original. (Wilss 1982: 62)

Many similar definitions can be found in the literature (cf. Koller 1979: 109111,
186; 1992: 8992; discussed in Pym 1992, 2004: 5759). These definitions would
all seem to cover the sort of thing that happens to the names of game-shows. You
go from one language to the other, and the result is a translation if and when a
relationship of equivalence is established on some level.
Look closely at the definitions. In each case, the term equivalent describes
one side only, the target side. The processes (replace, reproduce, lead) are
profoundly directional: translation goes from one side to the other, but not back
again. If we ask what the target-side equivalent is actually equivalent to, we find an
interesting array of answers: elements of a language, textual material, the mes-
sage, source-language text. The theories in this paradigm would seem to agree
on some things (target-side equivalents, directionality) but not on others (the na-
ture of the thing to translate).
278 Anthony Pym

In any theory, look for the definition of translation and try to see what it is as-
suming, then what it is omitting. What you find usually indicates the strengths and
weaknesses of the whole theory. In this case, the strength of the definitions is that
they have the one term (equivalent) that distinguishes translation from all the
other things that can be done in interlingual communication (rewriting, commen-
tary, summary, parody, etc.). The weakness is that they mostly do not explain why
this relation should just be one-way. Further, they are often in doubt as to whether
the equivalent is equal to a position or value within a language, a message, a text
with content and style, or to all those things but at different times.
We will describe those definitions as proposing a notion of directional
equivalence, at least to the extent that they forget to tell us about equivalence as an
affair of equal relations, or movements that could go either way. This might seem
like splitting hairs, but its importance will soon be clear.
Opposed to this one-way directionality, we also find notions of equivalence
that emphasize two-way movements. On this view, a relation of equivalence can
be tested by a simple test of back-translation. We can go from Friday to viernes
and then back to Friday, and it makes no difference which term is the source and
which the translation. We might term this two-way kind of equivalence natural,
at least in the sense that the correspondence existed in some way prior to the act
of translation (this is how Nida and Taber used the term natural in the definition
given above). On the level of bad luck, we could go from Friday 13th to martes y
13, and back again. The test might work for Le juste prix, and even for Der Preis ist
heiss, if we define carefully the levels we are operating on. But why does the French
apparently not have Le prix juste? And what about the lifelines that become
jokers and wild-cards? Can they also be justified as being in any way natural?
For that matter, what should we say about the Friday the 13th that is recognized
in Taiwan (we are told) not because it was always in the culture but because it trav-
eled there in the title of a horror film? Some kinds of equivalence refer to what is
done in a language prior to the intervention of the translator (hence the illusion
of the natural); others refer to what translators can do in the language (hence the
directionality of the result).
Directional and natural are terms that we are using here to describe the
different concepts used by theories of translation; they are not words used by the
theories themselves. They nevertheless help make some sense of a rather confus-
ing terrain. Most of the questions coming from structuralist linguistics concerned
strictly natural equivalence, or the vain search for them. When we mentioned Sau-
ssures sheep and mouton example, we talked about them translating each other.
The same would hold for Polish milk and universal bad-luck days. For that lin-
guistic paradigm, it should make no difference which of the terms is the source
and which is the target. For the above definitions of translation, on the other hand,
 Natural and directional equivalence in theories of translation 279

equivalence was something that resulted from a directional movement. They were
adopting quite a different approach to the concept.
The reference to directionality was perhaps the most profound way in which
the problem of structuralist linguistics was solved. To understand this, however,
we must first grasp the directions in which the naturalistic theories were heading.

5. Strategies for maintaining natural equivalence

One of the most entertaining texts in translation theory is the introduction to

Vinay and Darbelnets Stylistique compare du franais et de langlais, first pub-
lished in 1958. The two French linguists are driving from New York to Montral,
noting down the street signs along the way:
We soon reach the Canadian border, where the language of our forefathers is
music to our ears. The Canadian highway is built on the same principles as the
American one, except that its signs are bilingual. After SLOW, written on the
road-surface in enormous letters, comes LENTEMENT, which takes up the entire
width of the highway. What an unwieldy adverb! A pity French never made an
adverb just using the adjective LENT. But come to think of it, is LENTEMENT
really the equivalent of SLOW? We begin to have doubts, as one always does when
shifting from one language to another, when our SLIPPERY WHEN WET reap-
pears around a bend, followed by the French GLISSANT SI HUMIDE. Whoa!, as
the Lone Ranger would say, lets pause a while on this SOFT SHOULDER, thank-
fully caressed by no translation, and meditate on this SI, this if , more slippery
itself than an acre of ice. No monolingual speaker of French would ever have come
straight out with the phrase, nor would they have sprayed paint all over the road
for the sake of a long adverb ending in -MENT. Here we reach a key point, a sort
of turning lock between two languages. But of course parbleu! instead of
LENTEMENT [adverb, as in English] it should have been RALENTIR [verb in the
infinitive, as in France]! (1958: 19; our translation)

What kind of equivalence is being sought here? The kind the linguists actually
find, exemplified by the long French adverb lentement, is fair enough in directional
terms, since it says virtually the same thing as the English adverb slow. It changes
the length, but there is apparently room on the road. What worries the linguists
is that the sign Lentement is not what one would find on roads in France. For
them, the equivalent should be the verb Ralentir, since that is what would have
been used if no one had been translating from English (and as if Canada were
itself within France). This second kind of equivalence is thus non-directional, in
fact non-translational. It is what different languages and cultures seem to produce
from within their own systems. This is certainly natural equivalence: Slow should
give Ralentir, which should give Slow, and so on.
280 Anthony Pym

Natural equivalents do exist, but rarely in a state of untouched nature. They

are most frequently the stuff of terminology, of artificially standardized words that
are made to correspond to each other exactly. All specialized fields of knowledge
have their terminology. They are unnaturally creating natural equivalents all the
time. In Vinay and Darbelnet, however, the artificially imposed glossaries are to
be avoided where possible. On the contrary, they are seeking equivalents char-
acterized as natural precisely because they have supposedly developed without
interference from meddling linguists, translators, or other languages. In terms of
this naturalism, the best translations are found when you are not translating. We
use this mode of thought whenever we consult parallel texts.
In the late 1950s and 1960s, equivalence was often thought about in this way.
The problem was not primarily to show what the thing was or what one wanted
to do with it (Vinay and Darbelnet might have asked what words were best at
actually making Canadian drivers slow down). The problem was to describe ways
in which equivalence could be attained in all the situations where there were no
obvious natural equivalents.
Vinay and Darbelnet worked from examples to define seven general strategies
that could be used in this kind of translation. Not all the strategies would count as
good ways to produce natural equivalence. For example, they condoned the use of
loans and calques (necessarily directional) when there was no natural equivalent
available. They allowed for literal translation (here meaning fairly straightfor-
ward word-for-word), all the time stressing that its directionality may give un-
natural results. The strategies of key interest to Vinay and Darbelnet were trans-
position (where there is a switching of grammatical categories) and modulation
(where adjustments are made for different discursive conventions), since these
were the main ways in which linguistic changes could be made in order to keep
semantic sameness. The remaining two strategies concerned cultural adjustments:
correspondence (actually called quivalence in the French version) would use all
the corresponding proverbs and referents (like Friday the 13th), and adaptation
would then refer to different things with loosely equivalent cultural functions: cy-
cling is to the French what cricket is to the British, or baseball to the Americans
(or gardening is to the British what having lovers is to the Italians, we are told). In
all, Vinay and Darbelnets strategies range from the highly directional at one end
to the highly naturalistic at the other. They were thus able to recognize some kind
of continuity across both kinds of equivalence.
There are quite a few theories that list strategies like this. Vinay and Darbel-
nets work was inspired by Malblanc (1944), who compared French and German.
They in turn became the basis for Vzquez-Ayora (1977), who worked on Span-
ish and English. Different kinds of equivalence-maintaining strategies have been
described in a Russian tradition including Fedorov (1953), Shveitser (1987) and
 Natural and directional equivalence in theories of translation 281

Retsker (1974), and by the American Malone (1988), all usefully summarized in
Fawcett (1997). Some of these theorists are more accessible than others, but none
seems to have substantially extended the approach we find in Vinay and Darbel-
net. The only strategy we would really want to add to the above table is compensa-
tion, where an equivalent that cannot be found in one place or on one level in the
translation is made up for elsewhere. A source-text dialect, for example, might be
replaced by a few adjectives describing the speaker concerned, or functional use
of the intimate and formal second persons (a distinction that has been lost in Eng-
lish) might be compensated for by the selection of intimate or formal lexical items
(more examples can be found in Fawcett 1977).
The lists of strategies all make perfect sense when they are presented alongside
their carefully selected examples. However, when you get a translation and you try
to say exactly which strategy has been used where, you usually find that several
strategies explain the same equivalence, and some equivalence relations do not fit
comfortably into any. Vinay and Darbelnet recognize this problem:
The translation (on a door) of PRIVATE as DFENSE DENTRER [Prohibition
to Enter] is at once a transposition, a modulation and a correspondence. It is a
transposition because the adjective private is rendered by a noun phrase; it is a
modulation because the statement becomes a warning (cf. Wet Paint: Prenez garde
la peinture), and it is a correspondence because the translation has been pro-
duced by going back to the situation without bothering about the structure of the
English-language phrase. (1958: 54)

If three categories explain the one phenomenon, do we really need all the catego-
ries? Or are there potentially as many categories as there are equivalents?
The theories are rather vague about how natural equivalence works. They
mostly assume there is a piece of reality or thought (a referent, a function, a mes-
sage) that stands outside all languages and to which two languages can refer. The
thing would thus be a third element of comparison, a tertium comparationis, avail-
able to both sides. The translator thus goes from the source text to this thing, then
from the thing to the corresponding target text. Non-natural translations will re-
sult when one goes straight from the source text to the target text, as in the case of
Slow rendered as Lentement.
Perhaps the best-known account of this process is the one formulated by the
Parisian theorist Danica Seleskovitch. For her, a translation can only be natural if
the translator succeeds in forgetting entirely about the form of the source text. She
recommends listening to the sense, or deverbalizing the source text so that you
are only aware of the sense, which can be expressed in all languages. This is the
basis of what is known as the theory of sense (thorie du sens). From our historical
perspective, it is a process model of natural equivalence.
282 Anthony Pym

The great difficulty of this theory is that if a sense can be deverbalized, how
can we ever know what it is? As soon as we indicate it to someone, we have given it
a form of one kind or another. And there are no forms (not even the little pictures
or diagrams sometimes used) that can be considered truly universal. So there is
no real way of proving that such as thing as deverbalized sense exists. Listening
to the sense does no doubt describe a mental state that simultaneous interpret-
ers attain, but what they are hearing cannot be a sense without form. This theory
remains a loose metaphor with serious pedagogical virtues.
One of the paradoxes here is that process models like Seleskovitchs encour-
age translators not to look at linguistic forms in great detail, whereas the com-
parative methods espoused by Vinay and Darbelnet and the like were based on
close attention to linguistic forms in two languages. The process theories were
breaking with linguistics, tending to draw more on psychology (Seleskovitch
turned to the French psychologist Piaget). The comparative method, howev-
er, was entirely within linguistics. It would go on to compare not just isolated
phrases and collocations, but also pragmatic discourse conventions and modes
of text organization. Applied linguists like Hatim and Mason (1990 and 1997)
simply extend the level of comparison, generally remaining within the paradigm
of natural equivalence.
For the most idealistic natural equivalence, the ultimate aim is to find the pre-
translational equivalent that reproduces all aspects of the thing to be expressed.
Naturalistic approaches spend little time on defining translation; there is not much
analysis of different types of translation, or of translators having different aims.
Those things have somehow been decided by equivalence itself. Translation is sim-
ply translation. That is not always so, however, for directional equivalence.

6. Strategies for attaining directional equivalence

Questions about directional equivalence tend to concern what remains the same
and what is different after the transition from source to target. Most theories that
work within this sub-paradigm list not strategies, but different kinds of equiva-
lence. They also talk about different kinds of translating, which amounts to much
the same thing, since you translate quite differently depending on the level at
which you want equivalence to work.
Many of the theories here are based on just two types of equivalence, usually
presented as a straight dichotomy (you can translate one way or the other). Per-
haps the best known theorist of this kind is the American linguist Eugene Nida,
who argued that the Bible can be translated to achieve formal equivalence (fol-
lowing the words and textual patterns closely) or dynamic equivalence (trying
 Natural and directional equivalence in theories of translation 283

to recreate the function the words might have had in their original situation). For
example, the lamb of God that we know in English-language Christianity might
become the seal of God for an Inuit culture that knows a lot about seals but does
not have many lambs. That would be an extreme case of dynamic equivalence. On
the other hand, the name Bethlehem means House of Bread in Hebrew, so it
might be translated that way if we wanted to achieve dynamic equivalence on that
level. In that case, our translators traditionally opt for formal equivalence, even
when they use dynamic equivalence elsewhere in the same text. (Of course, things
are never quite that easy: the Arabic for Bethlehem, Beit Lahm, means House of
Meat so to whose name are we to be equivalent?)
As we have seen, Nidas definitions of translation claim to be seeking a natu-
ral equivalent. At one stage he toyed with Chomskys idea of kernel phrases as
the tertium comparationis. Nida, however, was mostly talking about translating
the Bible into the languages of cultures that are not traditionally Christian. What
natural equivalent should one find for the name of Jesus or God in a language
where they have never been mentioned? Whatever solution you find, it will prob-
ably concern a directional notion of equivalence, not a natural one. In this case,
an ideology of naturalness has been used to mask over the fact that the purpose of
translation is to change cultures.
A similar kind of dichotomy is found in the English translation critic Peter
Newmark (1981, 1988), who distinguishes between semantic and communica-
tive translation. The semantic kind of translation would look back to the formal
values of the source text and retain them as much as possible; the communicative
kind would look forward to the needs of the new addressee, and adapting to those
needs as much as necessary. Theories of directional equivalence mostly allow that
translators have to choose whether to render one aspect or another of the source
text. There is thus no necessary assumption of a natural equivalent.
For the Swiss theorist Werner Koller, whose German textbook on Translation
Studies went through four editions and many reprints between 1979 and 1992,
equivalents are what translators produce (cf. Pym 1997). In fact, equivalents do not
exist prior to the act of translation (cf. Stecconi 1994). Koller also shows that there
is no necessary restriction to just two kinds of equivalence. An equivalent can
be found for as many parts or levels of a source text as are considered pertinent.
Koller actually proposes five frames for equivalence relations: denotative (based on
extra-linguistic factors), connotative (based on way the source text is expressed),
text-normative (respecting or changing textual and linguistic norms), pragmatic
(with respect to the receiver of the target text) and formal (the formal-aesthetic
qualities of the source text). These categories suggest that the translator selects the
type of equivalence most appropriate to the dominant function of the source text.
The German theorist Katharina Rei (1971, 1976) was saying fairly similar things
284 Anthony Pym

in the same years. She recognizes three basic text types and argues that each type
requires that equivalence be sought on a different level.
There are important differences between the terms used in this set of theories.
Nidas two types of equivalence can potentially apply to any text whatsoever. The
same text could, for him, be translated in different ways for different audiences.
Koller and Rei, on the other hand, generally see the translators strategies as being
determined by the nature of the source text. The different usages of equivalence
are thus describing different things. Further, there would seem to be no strong rea-
son why there should be five ways to cut the cake (as in Koller), three (as in Rei)
or just two (as in Nida). There might be even more categories than those normally
considered in the theories, and many solutions that fall between the types.
Consider the problems of translating someones rsum or curriculum vitae.
Do you adapt the normal form of rsums in the target culture? Or do you just
reproduce that of the source culture? The solution is usually a mix, since the first
option means too much work, and the second option would mostly disadvantage
the person whose rsum it is. These days, however, most rsums are on a data-
base that can be printed out in several different formats and in several different
languages (English, Spanish and Catalan, in the case of our own university). The
results are somehow equivalent to something, but not in accordance with any of
the directional parameters listed above. In those cases, technology would seem to
have returned us to a natural equivalence of a particularly artificial kind. That is
where we are headed.

7. Equivalence as back-reference

Snell-Hornby, we noted, criticized the concept of equivalence as presenting an

illusion of symmetry between languages (Snell-Hornby 1988: 22). We are now in
a position to see that her criticism might be true of natural equivalence (especially
if tied to an ideology of natural usage), but it hardly holds for theories of direc-
tional equivalence. The naturalistic theories were basically analyzing languages,
battling within the paradigm of structuralist linguistics. Directional theories, on
the other hand, were working very much at the level of creative language use, in
keeping with attempts to analyze parole rather than langue.
Historically, the directional theories have been the most active within the
equivalence paradigm. This is partly because they encapsulate a dynamic view of
translation as a process. It is also because, when they come to analyze products
rather than processes, these theories can describe the way translations refer back
to their source texts. In this vein, the Czech theorist Ji Lev (1969: 32ff.) distin-
guished between anti-illusory and illusory translations. When you read an illu-
 Natural and directional equivalence in theories of translation 285

sory translation, you are not aware it is a translation; it has been so well adapted
to the target culture that it might as well be a text written anew. An anti-illusory
translation, on the other hand, retains some features of the source text, letting the
receiver know that it is a translation. This basic opposition has been reformulated
by a number of others. The German theorist Juliane House (1977, 1997) refers to
overt and covert translations. Christiane Nord (1988, 1997: 4752) prefers docu-
mentary and instrumental translations. The Israeli theorist Gideon Toury (1995)
talks about translations being adequate (to the source text) or appropriate (to the
circumstances of reception); the American theorist Lawrence Venuti (1995) op-
poses resistant to fluent translations. Lying behind all of these we find the early
nineteenth-century German preacher and translator Friedrich Schleiermacher
(1813) arguing that translations could be either foreignizing (verfremdend) or do-
mesticating (verdeutschend, Germanizing). Although these oppositions are all
saying slightly different things, they would all more or less fit in with Schleier-
machers description of two possible movements:
Either the translator leaves the author in peace, as much as possible, and moves
the reader toward that author, or the translator leaves the reader in peace, as much
as possible, and moves the author toward that reader.

As in Nida, here we are talking about a choice made by the translator, not necessar-
ily by the nature of the source text. What is strange here is that so many theorists
should be content with a simple either/or choice.
All these oppositions could be regarded as operating within the equivalence
paradigm. In all cases the two ways to translate can both lay claim to represent
some aspect or function of the source. The first term of each opposition would be
something like Nidas formal equivalence; the second term would incorporate
some degree of dynamic equivalence. So translation theorists would be saying
the same thing over and over, down through the centuries. Then again, try to apply
these oppositions to the simple examples at the beginning of this paper. If we take
martes y 13, we know that a formal translation would refer to Tuesday 13th, and
a dynamic-equivalence translation would give Friday 13th. Now, which of those
two translations is foreignizing? Which is domesticating? Which is moving the
reader? Which is moving the author? It seems impossible to say, at least until we
have a little more information. Or rather, both translations could be domesticating
in their way. If we wanted something foreignizing (anti-illusory, overt, documen-
tal, adequate, resistant) we would have to consider saying something like bad-luck
martes 13th, Tuesday 13th, bad-luck day, or even Tuesday 13th, bad-luck day in
Spanish-speaking countries. Is this kind of translation equivalent? Certainly not
on the level of form (in the last rendition we have added a whole phrase). Could
we claim equivalence in terms of function? Hardly. After all, a simple referential
286 Anthony Pym

phrase has become a whole cultural explanation, at a place where the source text
need offer no explanation. Some would say that the explanation is not equivalent,
since our version is too long to be a translation. Others might claim that this kind
of expansion is merely taking implicit cultural knowledge and making it explicit,
and since the cultural knowledge is the same, equivalence still reigns. Our version
might then be a very good translation.
This is a point at which natural equivalence breaks down. Directionality be-
comes clearly more important; we could use it to justify quite significant textual
expansion or reduction. The equivalence paradigm nevertheless tends to baulk at
this frontier. How much explanatory information could we insert and still claim
to be respecting equivalence? There is no clear agreement. The debate has become
about what is or is not a translation. And that is a question that the equivalence
paradigm was never really designed to address (it merely assumed an answer).

8. Only two categories?

Is there any reason why so many directional theories of equivalence have just two
categories? It seems you can translate just one way or the other, with not much in the
middle. However, many translation problems can be solved in more than two ways.
Naturalistic approaches can indeed have many more than two categories (Vinay and
Darbelnet, for example, listed seven main strategies). How should we explain this
profound binarism on the directional side? Let us just suggest two possibilities.
First, there may be something profoundly binary within equivalence-based
translation itself. To grasp this, translate the following sentence into a language
other than English (preferably not Dutch or German for this one!):
The first word of this very sentence has three letters.

In French this would give:

Le premier mot de cette phrase a trois lettres.

Here the word-level equivalence is fine, but functional equivalence has been lost.
A true self-reference has become a false self-reference, given that the first word of
the French sentence has two letters, not three (cf. the analysis of this example in
Burge 1978). How should the English sentence be translated? One might try the
following:
Le premier mot de cette phrase a deux lettres.

This tells us that the first word of the French sentence has two letters. We have lost
word-level equivalence with the English, but we have maintained the truth of the
 Natural and directional equivalence in theories of translation 287

self-reference. Our translation would seem to have moved from anti-illusory to il-
lusory, documentary to instrumental, adequate to appropriate, and the rest. In this
example, there would seem to be only these two possibilities available, one kind
of equivalence or the other. Or are there any further possibilities that we have not
yet considered?
A second possible reason for just two categories can be found in the early
nineteenth century. As we have seen, Friedrich Schleiermacher (1813) argued that
there were only two basic movements: you either move the author toward the
reader, or the reader toward the author. This is because just as they must belong to
one country, so people must adhere to one language or another, or they will wan-
der untethered in an unhappy middle ground (1813: 63). Translators, it seems,
cannot have it both ways.
If we look at these two reasons, they are both basically saying the same thing.
Translation has two sides (source and target), and thus two possible ways of achieving
self-reference, and two possible contexts from which the translator can speak. This
might suggest that directional equivalence is a particularly good mode of thought
for certain kinds of translation, and that those kinds, with just two basic sides, are
particularly good for keeping people on one side or the other, in separate languages
and countries. At the same time, is not all translation good at precisely that?

9. Relevance theory

Ernst-August Gutt (1991, second edition 2000) proposes a very elegant theory
that addresses these problems. Gutt looks at theories of natural equivalence (of the
kind we have seen from Vinay and Darbelnet, or Koller) and says that, in principle,
there is no limit to the kinds of equivalence that they can establish. Every text, in
fact every translation decision, would have to have its own theory of equivalence.
So all these theories are seriously flawed (a theory should have fewer terms than
the object it accounts for).
To overcome this difficulty, Gutt looks closely not at language or translations
as such, but at the kinds of things people believe about translations. Here he distin-
guishes between different kinds of translation, using two binary steps:
Overt translations are texts marked as translations, whereas covert trans-
lations would be things like the localization of a publicity campaign for a
new audience, which may as well not be a translation. Receivers of the co-
vert translation will not have any special beliefs about its equivalence or non-
equivalence, so Gutt is not interested in them.
Within the category of overt translations, considered to be translation proper,
there are two kinds. Indirect translation covers all the kinds of translations
288 Anthony Pym

that can be done without referring to the original context of the source text.
Direct translation would then be the kind that does refer to that context. In
Gutts terms, it creates a presumption of complete interpretative resemblance
(1991: 186). When we receive a direct translation, we think we understand
what receivers of the original understood, and that belief is not dependent on
any comparison of the linguistic details.
Here the critique of natural equivalence (too many possible categories) brings us
back to just the two categories (direct vs. indirect). And those two, we can
now see, are very typical of directional equivalence. That alone should be reason
enough for seeing Gutt as a theorist of equivalence.
What makes Gutts approach especially interesting here is the way he explains
directional equivalence as interpretative resemblance. He regards language as be-
ing a very weak representation of meaning, no more than a set of communica-
tive clues that receivers have to interpret. When he sets out to explain how such
interpretation is carried out, Gutt draws on the concept of inference, formulated
by the philosopher H. Paul Grice (1975). The basic idea here is that we do not com-
municate by language alone, but by the relation between language and context.
Consider the following example used by Gutt:
(1) Source text: Mary: The back door is open.
(2) Source context: If the back door is open, thieves can get in.
(3) Intended implicature: We should close the back door.

If we know about the context, we realize that the source text is a suggestion or
instruction, not just an observation. What is being said (the actual words of the
source text) is not what is being meant (the implicature produced by these words
interacting with a specific context). Grice explained such implicatures as operat-
ing by breaking various maxims, here the maxim Be relevant. If we know about
the context and the maxims, we can reach the implicature. If we do not, we will
not understand what is being said. Note that Grices maxims are not rules for pro-
ducing good utterances; they are rules that are regularly broken in order to pro-
duce implicatures. The actual maxims might thus vary enormously from culture
to culture. This variability is something that the British linguists Dan Sperber and
Deidre Wilson (1988) tend to sidestep when they reduce the Gricean analysis to
the one maxim Be relevant. They thus produced relevance theory, in fact say-
ing that all meaning is produced by the relation between language and context.
Implicatures are everywhere. And it is from relevance theory that Gutt developed
his account of translation.
If we were going to translate source text (1), we would have to know if the re-
ceiver of the translation has access to the context (2) and to the pragmatic maxim
 Natural and directional equivalence in theories of translation 289

being broken. If we can be sure of both kinds of access, we might just translate the
words of the text, producing something like formal equivalence. If not, we might
prefer to translate the implicature, somehow rendering the function, what the
words apparently mean. The notion of implicature might thus give us two kinds of
equivalence, in keeping with two kinds of translation.
Gutt, however, does not really want these two kinds of equivalence to be on
the same footing. He asks how Marys utterance should be reported (or translated).
There are at least two possibilities:
(4) Report 1: The back door is open.
(5) Report 2: We should close the back door.

Gutt points out that either of these reports will be successful if the receiver has
access to the source context; we may thus establish equivalence on either of those
levels. Two paths, not one. What happens, however, when the new receiver does
not have access to the source context? Let us say, we do not know about the pos-
sibility of thieves, and we are more interested in the children being able to get in
when they come home from school. If the reporter is working in this new context,
only the second report (5), the one that renders the implicature, is likely to be suc-
cessful. It will tell us that the back door should still be closed, even if there remain
doubts about the reason. Gutt, however, believes that direct translation should al-
ways allow interpretation in terms of the source context only. His preference would
be for the first report (4). For him, something along the lines of the second report
(5) would have no reason to be a translation.
Gutts application of relevance theory might be considered idiosyncratic on
this point. This could be because he has a particular concern with Bible transla-
tion. In insisting that interpretation should be in terms of the source context, Gutt
effectively discounts much of the dynamic equivalence that Eugene Nida wanted
to use to make biblical texts relevant to new audiences. Gutt insists not only that
the original context is the one that counts, but also that this makes the explication
of implicatures both unnecessary and undesirable (1991: 166). In the end, it is
the audiences responsibility to make up for such differences (ibid.). Make the re-
ceiver work! In terms of our example, the receiver of the second report (5) should
perhaps be smart enough to think about the thieves. Only when there is a risk of
misinterpretation should the translator inform the audience about contextual dif-
ferences, perhaps by adding, because there might be thieves.
At this point, the equivalence paradigm has become quite different from the
comparing of languages. The application of relevance theory shows equivalence to
be something that operates more on the level of beliefs, of fictions, or of possible
thought processes. It is thus something that can have consequences for the way
translators make decisions.
290 Anthony Pym

10. Equivalence as an illusion

With respect to equivalence (regardless of his personal translation preferences),

Gutt got it just about right. Translations, when they are accepted as such, do indeed
create a presumption of interpretative resemblance, and that presumption, no
matter how erroneous, may be all there ever was to equivalence. There is then no
need to go further; no need actually to test the pieces of language according to any
linguistic yardstick. Equivalence is always presumed equivalence, and no more.
In this, Gutts position is deceptively close to Toury (1980: 6370, 1995), where
all translations manifest equivalence simply because they are translations. The
work is then to analyze what the translations actually are (which is where equiva-
lence becomes a non-issue for Descriptive Translation Studies). Gutts location of
equivalence is also very much in tune with Pym (1992, 1995), except that Pym
stresses that the belief in equivalence is historical, shared, and cost-effective in
many situations: the translator is an equivalence producer, a professional com-
municator working for people who pay to believe that, on whatever level is perti-
nent, B is equivalent to A (1992: 77).
Gutt, Toury and Pym might thus fundamentally agree on equivalence as a
belief structure that has to be analyzed as such. Paradoxically, this kind of rough
consensus also logically marks the end of equivalence as a central concept. Thanks
to that agreement, linguists may venture into pragmatics, descriptive translation
scholars can collect and analyze translation shifts, and historians might want to
similarly shelve equivalence as an idea, operative only because of a conjuncture of
far more interesting sociological reasons. All those avenues take debate away from
equivalence itself. In so doing, they minimize the tussle between the natural and
the directional, killing the internal dynamics of the concept itself.
Equivalence might thus appear to be dead, except for the occasional decon-
structionist who has read little translation theory and is in need of a cheap Feind-
bild. Then again, history has not finished.

11. A problem not solved

Our purpose here has not been to support any hypothetical return to the equiva-
lence paradigm. We have instead sought to dispel some of the more frequent mis-
understandings associated with the term, notably the idea that equivalence means
domestication, or that it is opposed to creativity, or that it only comes in one flavor.
Indeed, understood in our terms, the concept may be able to address problems
quite different from those it was designed for.
 Natural and directional equivalence in theories of translation 291

As we struggle to see what the localization industry is doing to translation,

just a few things seem clear. First, the very term localization is being used to rob
translation of its more creative or adaptational aspects. When one talks about
the localization of software or websites, translation means the interlingual re-
placement of natural-language sentences or phrases, in tune with the narrowest of
linguistic approaches from the 1960s or 1970s. On the face of it, equivalence has
returned. Translation practice is being restricted to the kinds of decontextualized
examples used in the bad old days, except that now the decontextualization is not
just in the examples, it is a result of the technologies used in the practice itself. A
whole generation of translation theory has been undone.
Second, localization projects make use of all kinds of databases to keep trans-
lation within that narrow range of equivalence. Standard glossaries or read-only
term databases exemplify the conceptual geometries of natural equivalence. The
catch, of course, is that there is nothing natural about them. They tend to be im-
posed by the companies organizing or paying for the project. But natural equiva-
lence is precisely the mode of thought they make their translators adopt.
Third, when translation-memory tools are employed in these projects, the pro-
cess is very much in tune with directional equivalence. Equivalence is produced
from the languages of internationalization into the languages of end-use.
By thus applying the metalanguage of equivalence to a new kind of problem-
atic, some insight might result, and a few interesting questions can be asked. For
instance, if terminology tools fix natural equivalence and translation memories
fix directional equivalence, what kinds of movement can occur between the two?
When directional pairs are verified and enter the database, does not the result of
human decision become falsely naturalized? And when translators are obliged to
use read-only translation memories (such that they are unable to correct any per-
ceived mistakes in the memories), is not the directional activity subordinated to
the falsely natural? Finally, to test this entire illusion of controlled naturalness, can
the databases be used in reverse, to go from target language to source? The practice
is so rare as to be called reverse localization (Schler 2005).
The contemporary use of translation within localization projects thus retains
huge doses of equivalence, mostly of a highly retrograde kind. Gone is the kind of
re-creative equivalence that was once allowed for by the hunt for the natural nu-
ance; gone is the directional creativity by which translators consciously introduced
the new. Instead, equivalence returns to ensure the imposition of controlled pat-
terns on all cultures. And it does so at a time when the metalanguage of equiva-
lence has lost its exploratory and critical force.
Look closely, and you might still see the killers footprints.
292 Anthony Pym

Note

* The final version of this paper was written while the author was Visiting Research Fellow at
the University of Western Sydney, Australia, to which institution our thanks.

References

Burge, Tyler. 1978. Self-reference and translation. F. Guenther and M. Guenther-Reutter, eds.
Meaning and translation: Philosophical and linguistic approaches. London: Duckworth,
1978. 137153.
Catford, John C. 1965. A linguistic theory of translation: An essay in applied linguistics. London:
Oxford University Press.
Coseriu, Eugenio. 1978. Falsche und richtige Fragenstellungen in der bersetzungstheorie.
Lillebill Grhs, Gustav Korln and Bertil Malmberg, eds Theory and practice of translation.
Bern, Frankfurt a.M. and Las Vegas: Peter Lang, 1978. 1732.
Fawcett, Peter. 1997. Translation and language: Linguistic theories explained. Manchester: St Je-
rome.
Fedorov, Andrei V. 1953. Vvedenie b teoriu perevoda [Introduction to the theory of translation].
Moscow: Literaturi na inostrannix iazikax.
Grice, H. Paul. 1975. Logic and conversation. Peter Cole and Jerry L. Morgan, eds Syntax and
semantics 3: Speech acts. New York: Academic Press, 1975. 4158.
Gutt, Ernst-August. 1991. Translation and relevance: Cognition and context. Oxford: Basil Black-
well. 2nd edition. Manchester: St Jerome, 2000.
Hatim, Basil and Ian Mason. 1990. Discourse and the translator. London: Longman.
Hatim, Basil and Ian Mason. 1997. The Translator as communicator. London and New York:
Routledge.
House, Juliane. 1977. A model for translation quality assessment. Tbingen: TBL Verlag Gunter
Narr.
House, Juliane. 1997. Translation quality assessment: A model revisited. Tbingen: Gunter Narr.
Koller, Werner. 1979. Einfhrung in die bersetzungswissenschaft. Heidelberg: Quelle & Meyer.
Koller, Werner. 1992. Einfhring in die bersetzungswissenschaft. 4., vllig neu bearbeitete Aufla-
ge. Heidelberg & Wiesbaden: Quelle & Meyer.
Lev, Ji. 1969. Die literarische bersetzung: Theorie einer Kunstgattung. Frankfurt/Main: Athe-
num.
Malblanc, Andr. 1963 Stylistique compare du franais et de lallemand: Essai de reprsentation
linguistique compare et tude de traduction. Paris: Didier.
Malone, Joseph L. 1988. The science of linguistics in the art of translation. Albany: State Univer-
sity of New York Press.
Mounin, Georges. 1963. Les Problmes thoriques de la traduction. Paris: Gallimard.
Newmark, Peter. 1981. Approaches to translation. Oxford: Pergamon Press.
Newmark, Peter. 1988. A textbook of translation. London: Prentice Hall.
Nida, Eugene A. 1959. Principles of translating as exemplified by Bible translating. Language
structure and thought: Essays by Eugene A. Nida, ed. Anwar S. Dil. Stanford: Stanford Uni-
versity Press, 1975.
 Natural and directional equivalence in theories of translation 293

Nida, Eugene A. and Charles R. Taber. 1969. The theory and practice of translation. Leiden: E.J.
Brill.
Nord, Christiane. 1988. Textanalyse und bersetzen. Heidelberg: Julius Groos.
Nord, Christiane. 1997. Translating as a purposeful activity: Functionalist approaches explained.
Manchester: St Jerome.
Oettinger, Anthony G. 1960. Automatic language translation: Lexical and technical aspects, with
particular reference to Russian. Cambridge Mass.: Harvard University Press.
Pym, Anthony. 1992. Translation and text transfer: An Essay on the principles of intercultural
communication. Frankfurt am Main: Peter Lang.
Pym, Anthony. 1995. European Translation Studies, une science qui drange, and why equiva-
lence neednt be a dirty word. TTR 8:1. 153176.
Pym, Anthony. 1997. Kollers quivalenz revisited. The translator 3:1. 7179.
Pym, Anthony. 2004. The moving text: Translation, localization, and distribution. Amsterdam
and Philadelphia: John Benjamins.
Rei, Katharina. 1971. Mglichkeiten und Grenzen der bersetzungskritik. Mnchen: Max Hue-
ber. [Hueber Hochschulreihe 12.]
Rei, Katharina. 1976. Texttypen, bersetzungstypen und die Beurteilung von bersetzun-
gen. Lebende Sprachen 22:3. 97100.
Retsker, Yakob I. 1974. Teoria perevoda i perevodcheskaia praktika [Theory of and translation
practice]. Moscow: Mezhdunarodnii otnoshenia.
Saussure, Ferdinand de. 1916. Cours de linguistique gnrale, eds. C. Bally and A. Sechehaye. Tr.
W. Baskin as Course in general linguistics. Glasgow: Fontana Collins, 1974.
Schler, Reinhard. 2005. Reverse localisation. Paper presented to the Aslib conference Trans-
lating and the computer 27, November 24, 2005.
Schleiermacher, Friedrich. 1813. Ueber die verschiedenen Methoden des Uebersezens. Reprin-
ted in Hans Joachim Strig, ed. Das Problem des bersetzens. Darmstadt: Wissenschaft
liche Buchgesellschaft, 1963. 3870. Tr. Douglas Robinson as On the different methods of
translating. Douglas Robinson, ed. Western translation theory from Herodotus to Nietzsche.
Manchester: St Jerome, 1997. 225238.
Shveitser, Aleksandr D. 1987. bersetzung und Linguistik. Berlin: Akademie Verlag. Original
title: Perevod i Lingvistika, 1973.
Snell-Hornby, Mary. 1988. Translation Studies. An integrated approach. Amsterdam and Phila-
delphia: John Benjamins.
Sperber, Dan and Deirdre Wilson. 1988. Relevance: Communication and cognition. Cambridge
Mass.: Harvard University Press.
Stecconi, Ubaldo. 1994. Peirces semiotics for translation. Yves Gambier and Mary Snell-Horn-
by, eds. Problemi e tendenze nella didattica dell interpretazione e della traduzione / Problems
and trends in the teaching of interpreting and translation. Koin 4 (Misano Adriatico), 1994.
161180.
Toury, Gideon. 1980. In search of a theory of translation. Tel Aviv: Porter Institute for Poetics
and Semiotics.
Toury, Gideon. 1995. Descriptive Translation Studies and beyond. Amsterdam and Philadelphia:
John Benjamins.
Vzquez-Ayora, Gerardo. 1977. Introduccin a la traductologa. Washington DC: Georgetown
University Press.
Venuti, Lawrence. 1995. The translators invisibility. A history of translation. London and New
York: Routledge.
294 Anthony Pym

Vinay, Jean-Paul and Jean Darbelnet. 1958. Stylistique compare du franais et de langlais: m-
thode de traduction. Nouvelle dition revue et corrige, 1972. Paris: Didier.
Wilss, Wolfram. 1982. The science of translation: Problems and Methods. Tbingen: Gunter
Narr.

Rsum

Le concept dquivalence a t la clef de vote des thories linguistiques de la traduction depuis

les annes 1960, bien que sa gomtrie conceptuelle puisse remonter jusqu Cicron. Vues de
prs, quelques-unes de ces thories prsupposent des quivalents qui existent bien avant le mo-
ment de la traduction, dont lhorizon idologique serait alors lquivalence naturelle. Dautres
thories, pourtant, projettent un traducteur plutt crateur des quivalences, dont lhorizon se-
rait alors aussi directionnel que lest lacte de traduire. Lquivalence naturelle concerne lillu-
sion des langues prtraductionnelles; lquivalence directionnelle recherche la productivit des
langues dans et grce au passage traductionnel. Lentrelacement de ces deux penses a donn lieu
maints malentendus, voire des critiques injustes du concept dquivalence. La fin historique
du concept est pourtant survenue au moment o lon a conceptualis lquivalence direction-
nelle comme croyance ou attente du ct du rcepteur, sans besoin de vrification en termes
linguistiques. En mme temps, la communication lctronique rend moins stables les textes de
dpart, ce qui rend plus difficile lquivalence comme fidelit une valeur fixe, tout comme les
relations entre les langues deviennent plus hirarchiques, ce qui brise les illusions de lquivalen-
ce naturelle. Plus caduc que jamais, le mtalangage de lquivalence pourrait nanmoins jouer
un rle clef dans lanalyse critique du discours et de la pratique de la localisation contemporaine,
o les technologies de la terminologie et des mmoires de traduction imposent respectivement
au traducteur, sans les termes, lquivalence naturelle et lquivalence directionnelle.

Authors address
Anthony Pym
Universitat Rovira i Virgili
Plaa Imperial Trraco 1
E-43005 Tarragona
Spain
e-mail: [email protected]