N O R M S IN T R A N S L A T I O N 199
C h a p t e r 16
Gideon Toury
O W E V E R H IG H L Y O N E m ay th ink o f L in gu istics, T ex t-L in g uistics,
Contrastive Textology or Pragm atics and of their explanatory pow er with
H
respect to translation al phenom ena, being a translator cannot be reduced to the
mere generation of utterances which would be considered “ tran slation s” within
any o f these disciplines. Translation activities should rather be regarded as having
cultural significance. Consequently, “ translatorsh ip” am ounts first and foremost to
being able to play a social role, i.e., to fulfil a function allotted by a com munity—
to the activity, its practitioners and/or their products— in a w ay which is deemed
appropriate in its own terms o f reference. The acquisition o f a set o f norm s for
determining the suitability of that kind of behaviour, and for m anoeuvring between
all the factors which m ay constrain it, is therefore a prerequisite for becoming a
translator within a cultural environment.
The process by which a bilingual speaker m ay be said to gain recognition in his/
her capacity as a tran slator has hardly been studied so far. [...] In the present
chapter the nature o f the acquired norm s themselves will be addressed, along with
their role in directing translation activity in socio-culturally relevant settings. This
presentation will be followed by a brief discussion o f translation al norm s as a
second-order object of Translation Studies, to be reconstructed and studied within
the kind o f fram ew ork which we are now in the process o f sketching. As strictly
translational norms can only be applied at the receiving end, establishing them is
not merely justified by a target-oriented approach but should be seen as its very
epitome.
1978/revised 1995
1 R u les, n o rm s, id iosy n crasies
In its socio-cultural dimension, translation can be described as subject to constraints
o f several types and varying degree. These extend far beyond the source text; the
systemic differences between the languages and textual traditions involved in the
act, or even the possibilities and lim itations o f the cognitive apparatu s o f the
translator as a necessary mediator. In fact, cognition itself is influenced, probably
even m odified by socio-cultural factors. At any rate, translators perform ing under
different conditions (e.g., translating texts o f different kinds, and/or for different
audiences) often adopt different strategies, and ultimately com e up with m arkedly
different products. Something has obviously changed here, and I very m uch doubt
it that it is the cognitive apparatus as such.
In terms o f their potency, socio-cultural constraints have been described along a
scale anchored between two extrem es: general, relatively absolute rules, on the one
hand and pure idiosyncrasies on the other. Between these tw o poles lies a vast
middle-ground occupied by inter subjective factors com m only designated norm s.
The norm s themselves form a graded continuum along the scale: som e are stronger,
and hence more rule-like, others are weaker, and hence alm ost idiosyncratic. The
borderlines between the various types of constraints are thus diffuse. Each of the
concepts, including the grading itself, is relative to o. Thus w hat is just a favoured
mode o f behaviour within a heterogeneous group m ay well acquire much m ore
binding force within a certain (more hom ogeneous) section thereof, in term s of
either hum an agents (e.g., tran slators am ong texters in general) or types of activity
(e.g., interpreting, or legal translation, within translation at large).
A long the tem poral axis, each type o f constraint may, and often does m ove into
its neighbouring dom ain(s) through processes of rise and decline. Thus, mere, whims
may catch on and become m ore and m ore norm ative, and norm s can gain so much
validity that, for all practical purposes, they become as binding as rules; or the
other way around, o f course. Shifts o f validity and force often have to do with
changes of status within a society. In fact, they can always be described in connection
with the notion of norm , especially since, as the process goes on, they are likely to
cross its realm, i.e., actually become norm s. The other tw o types o f constraints
may even be redefined in terms of norms: rules as “ [more] objective” , idiosyncrasies
as “ [more] subjective [or: less inter subjective]” norm s.
S o cio lo g ists and social p sych o lo g ists have lon g re gard e d n orm s as the
translation of general values or ideas shared by a com m unity— as to w hat is
right and w rong, adeq u ate and in ad eq u ate— into perform ance in structions
ap p ro p riate for an d ap plicab le to p articu lar situ atio n s, specifying w hat is
prescribed and forbidden as well as w hat is tolerated and perm itted in a certain
behavioural dimension (the fam ous “ square o f norm ativity” , w hich has lately
been elaborated on with regard to translation in De G eest 19 9 2:3 8 40). N orm s
are acquired by the individual during his/her socialization and alw ays imply
s a n c tio n s — ac tu al or p o t en tial, n e g ativ e as w ell as p o s itiv e . W ithin the
community, norms also serve as criteria according to which actual instances of
behaviour are evaluated. Obviously, there is a point in assum ing the existence of
norm s only in situations which allow for different kinds o f behaviour, on the
additional condition that selection am ong them be nonrandom .1 Inasm uch as a
200 GIDEON TOURY
norm is really active and effective, one can therefore distinguish regularity of
behaviour in recurrent situations o f the same type, which w ould render regularities
a m ain source for any study o f norm s as well.
The centrality o f the norm s is not only m etaphorical, then, in terms of their
relative position along a postulated continuum o f constraints; rather, it is essential:
N orm s are the key concept and focal point in any attem pt to account for the social
relevance o f activities, because their existence, and the wide range o f situations
they apply to (with the conform ity this im plies), are the m ain factors ensuring the
establishment and retention of social order. This holds for cultures too, or for any of
the system s constituting them, which are, after all, social institutions ipso facto. O f
course, behaviour which does not conform to prevailing norm s is alw ays possible
too. M oreover, “ non-compliance with a norm in particular instances does not
invalidate the norm ” (Herm ans 1991:162). At the same time, there w ould norm ally
be a price to pay for opting for any deviant kind o f behaviour.
One thing to bear in mind, when setting out to study norm -governed behaviour,
is th at there is no necessary identity between the norm s themselves and any
form ulation of them in language. Verbal form ulations o f course reflect awareness
o f the existence o f norms as well as o f their respective significance. However, they
also imply other interests, particularly a desire to control behaviour i.e., to dictate
norm s rather than merely account for them. N orm ative form ulations tend to be
slanted, then, and should always be taken with a grain of salt.
2 T ran slatio n as a norm -gov erned activity
Translation is a kind o f activity which inevitably involves at least tw o languages
and tw o cultural traditions, i.e., at least tw o sets of norm -systems on each level.
Thus, the “ value” behind it m ay be described as consisting o f tw o m ajor elements:
1 being a text in a certain language, and hence occupying a position, or filling
in a slot, in the appropriate culture, or in a certain section thereof;
2 constituting a representation in that language/culture o f another, preexisting
text in som e other language, belonging to some other culture and occupying
a definite position within it.
These two types of requirement derive from two sources which— even though the
distance between them m ay vary greatly— are nevertheless alw ays different and
therefore often incom patible. Were it not for the regulative capacity of norm s, the
tensions between the tw o sources o f constraints w ould have to be resolved on an
entirely individual basis, and with no clear yardstick to go by. Extrem e free variation
may well have been the result, which it certainly is not. Rather, translation behaviour
within a culture tends to m anifest certain regularities, one consequence being that
even if they are unable to account for deviations in any explicit way, the persons-in-
the-culture can often tell when a translator has failed to adhere to sanctioned
practices.
It has proven useful and enlightening to regard the basic choice which can be
m ade between requirements of the tw o different sources as constituting an initial
N O R M S IN T R A N S L A T I O N 201
norm . Thus, a translator m ay subject him -/herself either to the original text, with
the norms it has realized, or to the norm s active in the target culture, or, in that
section o f it which w ould host the end product. If the first stance is adopted, the
translation will tend to subscribe to the norm s of the source text, and through
them also to the norms of the source language and culture. This tendency; which
has often been characterized as the pursuit o f adequate translation,2 m ay well
entail certain incom patibilities with target norm s and practices, especially those
lying beyond the mere linguistic ones. If, on the other hand, the second stance is
adopted, norm s system s o f the target culture are triggered and set into m otion.
Shifts from the source text w ould be an alm ost inevitable price. Thus, w hereas
adherence to source norm s determines a translation ’s adequacy as com pared to
the source text, subscription to norm s originating in the target culture determines
its acceptability.
Obviously, even the m ost adequacy-oriented translation involves shifts from the
source text. In fact, the occurrence o f shifts has long been acknow ledged as a true
universal o f translation. However, since the need itself to deviate from sour ce-text
patterns can alw ays be realized in m ore than one way, the actual realization of so-
called ob ligato ry sh ifts, to the extent th at it is n on-random , and hence not
idiosyncratic, is already truly nor m -governed. So is everything that has to do with
non-obligatory shifts, which are o f course m ore than just possible in real-life
translation: they occur everywhere and tend to constitute the m ajority o f shifting in
any single act o f human translation, rendering the latter a contributing factor to, as
well as the epitome of regularity.
The term “ initial n orm ” should not be overinterpreted, however. Its initiality
derives from its superordinance over particular norm s which pertain to lower, and
therefore more specific levels. The kind of priority postulated here is basically
logical, and need n ot coincide with any “ re al” , i.e., ch ro no logical order of
application. The notion is thus designed to serve first and forem ost as an explanatory
to o l. Even if no clear m acro-level tendency can be shown, any micro-level decision
can still be accounted for in terms of adequacy vs. acceptability. On the other hand,
in cases where an overall choice has been m ade, it is not necessary that every single
lower-level decision be m ade in full accord with it. We are still talking regularities,
then, but not necessarily of any absolute type. It is unrealistic to expect absolute
regularities anyway, in any behavioural dom ain.
A ctual tran slatio n decisio ns (the results o f w hich the researcher w ould
confront) will necessarily involve som e ad hoc com bination of, or com prom ise
between the two extrem es im plied by the initial norm . Still, for theoretical and
m ethodological reasons, it seems wiser to retain the opposition and treat the two
p oles as distin ct in prin ciple: If they are n ot re g ard ed as h av in g d istin ct
theoretical statuses, how w ould com prom ises differing in type or in extent be
distinguished and accounted for?
Finally, the claim that it is basically a norm -governed type o f behaviour applies
to translation o f all kinds, not only literary, philosoph ical or biblical translation,
which is where m ost norm -oriented studies have been conducted so far. As has
recently been claim ed and dem onstrated in an all to o sketchy exchange o f views in
Target (M .Shlesinger 1989 and H arris 1990 ), sim ilar things can even be said of
conference interpreting. N eedless to say, this does not m ean that the exact sam e
202 GIDEON TOURY
conditions apply to all kinds o f translation. In fact, their application in different
cultural sectors is precisely one o f the aspects that should be subm itted to study. In
principle, the claim is also valid for every society and historical period, thus offering
a fram e w ork fo r h istorically orien ted stud ies w hich w ou ld also allow for
comparison.
3 T ran slatio n n orm s: an overview
N orm s can be expected to operate not only in translation o f all kinds, but also at
every stage in the translating event, and hence to be reflected on every level o f its
product. It has proven convenient to first distinguish two larger groups of norm s
applicable to translation: prelim inary vs. operational.
Preliminary norm s have to do with tw o m ain sets of considerations which are
often interconnected: those regarding the existence and actual nature of a definite
translation policy, and those related to the directness of translation.
Translation policy refers to those factors that govern the choice o f text types;
or even o f individual texts, to be im ported through translation into a particular
culture/language at a particular point in time. Such a policy will be said to exist
inasmuch as the choice is found to be non-random . Different policies m ay of
course apply to different subgrou ps, in terms o f either text-types (e.g. literary vs.
non-literary) or hum an agents and groups thereof (e.g., different publishing
houses), and the interface between the tw o often offers very fertile grounds for
policy hunting.
Considerations concerning directness o f translation involve the threshold of
tolerance for translating from languages other than the ultimate source language:
is indirect translation perm itted at all? In translating from w hat source languages/
text-types/periods (etc.) is it permitted/prohibited/tolerated/preferred? W hat are the
permitted/prohibited/tolerated/preferred m ediating languages? Is there a tendency/
obligation to m ark a translated work as having been mediated or is this fact ignored/
camouflaged/denied? If it is m entioned, is the identity o f the m ediating language
supplied as well? A nd so on.
O peration al norm s, in turn, m ay be conceived of as directing the decisions
m ade during the act of translation itself. They affect the m atrix o f the text— i.e.
the m odes o f distributing linguistic m aterial in it— as well as the textual m ake
up and verbal form ulation as such. They thus govern— directly or indirectly—
the relationships as well that w ould obtain between the target and source texts,
i.e., what is m ore likely to rem ain invariant under transform ation and w hat will
change.
So-called m atricial norm s m ay govern the very existence o f target-language
m aterial intended as a substitute for the corresponding source-language m aterial
(and hence the degree of fullness of translation), its location in the text (or the form
o f actual distribution), as well as the textual segm entation.3 The extent to which
om issions, additions, changes of location and m anipulations o f segmentation are
referred to in the translated texts (or around them) m ay also be determined by
norm s, even though the one can very well occur without the other.
Obviously, the borderlines between the various m atricial phenomena are not
N O R M S IN T R A N S L A T I O N 2 03
clear-cut. For instance, large-scale om issions often entail changes o f segm entation
as well, especially if the om itted portions have no clear boundaries, or textual-
linguistic standing, i.e., if they are not integral sentences, paragraphs or chapters.
By the sam e token, a change o f location m ay often be accounted for as an om ission
(in one place) com pensated by an addition (elsewhere). The decision as to what
may have “ really” taken place is thus description-bound: W hat one is after is (more
or less cogent) explanatory hypotheses, not necessarily “ true-to-life” accounts, which
one can never be sure o f anyway.
T extual-linguistic norm s, in turn, govern the selection of m aterial to form ulate
the target text in, or replace the original textual and linguistic material with. Textual-
linguistic norm s m ay either be gen eral, and hence apply to tran slatio n qua
translation, or particular, in which case they w ould pertain to a particular text-type
and/or m ode o f translation only. Som e o f them m ay be identical to the norm s
governing non-translational text-production, but such an identity should never be
taken for granted. This is the m ethodological reason why no study o f translation
can, or should proceed from the assum ption that the later is representative of the
target language, or of any overall textual tradition thereof. (And see our discussion
o f “ translation-specific lexical item s” .)
It is clear that preliminary norm s have both logical and chronological precedence
over the operational ones. This is not to say that between the tw o m ajor groups
there are no relationships whatsoever, including m utual influences or even tw o-
way conditioning. However, these relations are by no m eans fixed and given, and
their establishment form s an inseparable part of any study of translation as a norm-
governed activity. N evertheless, we can safely assum e at least that the relations
which do exist have to do with the initial norm . They m ight even be found to
intersect it— another im portant reason to retain the opposition between “ adequacy”
and “ acceptability” as a basic coordinate system for the form ulation of explanatory
hypotheses.4
O pe ratio n al norm s as such m ay be describe d as serving as a m ode l, in
accordance with which translations com e into being, w hether involving the norm s
realized by the source text (i.e., adequate translation) plus certain m odifications or
purely target norms, or a particular com prom ise between the tw o. Every model
supplying perform ance instructions m ay be said to act as a restricting factor: it
opens up certain options while closing others. Consequently, when the first position
is fully adopted, the translation can hardly be said to have been m ade into the
target language as a whole. Rather, it is m ade into a m odel language, which is at
best som e part o f the former and at w orst an artificial, and as such nonexistent
variety.5 In this last case, the translation is not really introduced into the target
culture either, but is im posed on it, so to speak. Sure, it m ay eventually carve a
niche for itself in the latter, but there is no initial attem pt to accom m odate it to any
existing “ slot” . O n the other hand, when the second position is adopted, what a
translator is introducing into the target culture (which is indeed w hat s/he can be
described as doing now) is a version of the original w ork, cut to the m easure o f a
preexisting model. (And see our discussion of the opposition between the “ translation
o f literary te xts” and “ literary translation ” as well as the detailed presentation of
the H ebrew translation o f a G erm an Sch laraffenland text.)
204 GIDEON TOURY
The apparent contradiction between any traditional concept of equivalence and
the limited model into which a translation has just been claim ed to be m oulded can
only be resolved by postulating that it is norm s that determ ine the (type and extent
of) equivalence manifested by actual translations. The study of norms thus constitutes
a vital step tow ards establishing just how the functional-relational postulate of
equivalence has been realized— whether in one translated text, in the w ork o f a single
translator or “ school” o f translators, in a given historical period, or in any other
justifiable selection.6 W hat this approach entails is a clear wish to retain the notion
o f equivalence, which various contemporary approaches (e.g. H onig and Kussm aul
1982; H olz-M ânttâri 1984; Snell-H ornby 1988) have tried to do without, while
introducing one essential change into it: from an ahistorical, largely prescriptive
concept to a historical one. Rather than being a single relationship, denoting a
recurring type o f invariant, it com es to refer to any relation which is found to have
characterized translation under a specified set of circum stances.
At the end o f a full-fledged study it will probably be found that translational
norm s, hence the realization o f the equivalence postulate, are all, to a large extent,
dependent on the position held by translation— the activity as well as its products—
in the target culture. An interesting field for study is therefore com parative: the
nature o f translational norm s as com pared to those governing non-translational
kinds o f text-production. In fact, this kind of study is absolutely vital, if translating
and translations are to be appropriately contextualized.
4 T h e m u ltiplicity o f tran slatio n al norm s
The difficulties involved in any attem pt to account for tran slational norm s should
not be underestimated. These, however, lie first and foremost in tw o features inherent
in the very notion o f norm , and are therefore not unique to Translation Studies at
all: the socio-cultural specificity o f norm s and their basic instability.
Thus, whatever its exact content, there is absolutely no need for a norm to
apply— to the sam e extent, or at all— to all sectors within a society. Even less
necessary, or indeed likely, is it for a norm to apply across cultures. In fact,
“ sam en ess” here is a mere coincidence— or else the result o f continuous contacts
between subsystems within a culture, or between entire cultural systems, and hence
a m anifestation of interference. (For some general rules o f systemic interference see
Even-Zohar 1 99 0 :53 -72 .) Even then, it is often m ore a matter of apparent than of
a genuine identity. After all, significance is only attributed to a norm by the system
in which it is em bedded, and the system s rem ain different even if instances of
external behaviour appear the same.
In addition to their inherent specificity, norm s are also unstable, changing
entities; not because o f any intrinsic flaw but by their very nature as norms. At
times, norm s change rather quickly; at other tim es, they are m ore enduring, and
the process m ay take longer. Either way, substantial changes, in translational norm s
too, quite often occur within one’s life-time.
O f course it is not as if all translators are passive in face o f these changes.
Rather, many o f them, through their very activity, help in shaping the process, as
N O R M S IN T R A N S L A T I O N 2 05
do translation criticism , translation ideology (including the one em anating from
contem porary academe, often in the guise o f theory), and, o f course, various norm -
setting activities o f institutes where, in many societies, translators are now being
trained. W ittingly or unwittingly, they all try to interfere with the “ n atural” course
o f events and to divert it according to their own preferences. Yet the success o f their
endeavours is never fully foreseeable. In fact, the relative role o f different agents in
the overall dynamics o f translational norm s is still largely a m atter o f conjecture
even for times past, and much m ore research is needed to clarify it.
C om plying with social pressures to con stan tly a dju st one ’s behaviour to
norm s that keep changing is o f course far from sim ple, and m ost people—
including translators, initiators o f translation activities and the consum ers of
their products— do so only up to a point. Therefore, it is not all that rare to find
side by side in a society three types o f com peting norm s, each having its own
follow ers and a position of its own in the culture at large: the ones th at dom inate
the centre o f the system, and hence direct translational behaviour o f the so-called
m ainstream , alongside the rem nants o f previous sets o f norm s and the rudim ents
o f new ones, hovering in the periphery. This is why it is possible to speak— and
n ot d e ro g ato rily— o f being “ tre n d y” , “ o ld -fash io n e d ” or “ p ro g re ssiv e ” in
translation (or in any single section thereof) as it is in any other behavioural
dom ain.
O ne’s status as a translator m ay o f course be temporary, especially if one fails to
adjust to the changing requirem ents, or does so to an extent which is deemed
insufficient. Thus, as changes o f norm s occur, formerly “ progressive” translators
may soon find themselves just “ trendy” , or on occasion as even downright “p assé ".
At the sam e tim e, regarding this process as involving a m ere alternation of
generations can be misleading, especially if generations are directly equated with
age groups. While there often are correlations between one’s position along the
“ d ated ” — “ m ainstream ” — “ avant-garde” axis and one’s age, these cannot, and
should not be taken as inevitable, much less as a starting point and fram ew ork for
the study o f norm s in action. M o st notably, young people who are in the early
phases o f their initiation as tran slators often behave in an extremely epigonic way:
they tend to perform according to dated, but still existing norm s, the m ore so if they
receive reinforcem ent from agents h olding to dated norm s, be they language
teachers, editors, or even teachers o f translation.
M ultiplicity and variation should not be taken to im ply that there is no such
thing as norms active in translation. They only m ean that real-life situations tend
to be complex; and this com plexity had better be noted rather than ignored, if one
is to draw any justifiable conclusions. As already argued, the only viable w ay out
seems to be to contextualize every phenom enon, every item, every text, every act,
on the way to allotting the different norm s themselves their appropriate position
and valence. This is why it is sim ply unthinkable, from the point of view o f the
study o f translation as a norm -governed activity, for all item s to be treated on a
par, as if they were o f the sam e systemic position, the sam e significance, the sam e
level o f representativeness o f the target culture and its constraints. Unfortunately,
such an indiscrim inate approach has been all too com m on, and has often led to a
complete blurring of the norm ative picture, som etim es even to the absurd claim
that no norms could be detected at all. The only way to keep that picture in focus is
206 GIDEON TOURY
to go beyond the establishment o f mere “ check-lists” of factors which may occur in
a corpus and have the lists ordered, for instance with respect to the status o f those
factors as characterizing “ m ain stream ” , “ d ated” and “ avant-garde” activities,
respectively.
This imm ediately suggests a further axis of contextualization, whose necessity
has so far only been implied; namely, the historical one. After all, a norm can only
be m arked as “ date d” if it w as active in a previous period, and if, at that time, it
had a different, “ non-dated” position. By the same token, norm-governed behaviour
can prove to have been “ avant-garde” only in view o f subsequent attitudes towards
it: an idiosyncrasy which never evolved into som ething m ore general can only be
described as a norm by extension, so to speak (see Section 1 above). Finally, there
is nothing inherently “ m ainstream ” about m ainstream behaviour, except when it
happens to function as such, which m eans that it too is time-bound. W hat I am
claim ing here, in fact, is that historical contextualization is a must not only for a
diachronic study, which nobody w ould contest, but also for synchronic studies,
which still seems a lot less obvious unless one has accepted the principles o f so-
called “ D ynamic Functionalism ” (for which, see the Introduction to Even-Zohar
19907 and Sheffy 1992: passim ).
Finally, in translation too, non-normative behaviour is alw ays a possibility. The
price for selecting this option m ay be as low as a (culturally determined) need to
submit the end product to revision. However, it m ay also be far m ore severe to the
point o f taking aw ay one’s earned recognition as a translator; which is precisely
why non-norm ative behaviour tends to be the exception, in actual practice. O n the
other hand, in retrospect, deviant instances o f behaviour m ay be found to have
effected changes in the very system. This is why they constitute an im portant field
o f study, as long as they are regarded as what they have really been and are not put
indiscriminately into one basket with all the rest. Implied are intriguing questions
such as who is “ allow ed” by a culture to introduce changes and under what
circum stances such changes m ay be expected to occur and/or be accepted.
5 S tud ying tran slatio n al n orm s
So far we have discussed norms mainly in terms of their activity during a translation
event and their effectiveness in the act o f translation itself. To be sure, this is precisely
where and when translational norm s are active. However, what is actually available
for observation is not so much the norm s themselves, but rather norm-governed
instances of behaviour. To be even m ore precise, m ore often than not, it is the
products o f such behaviour. Thus, even when translating is claim ed to be studied
directly, as is the case with the use o f “ Thinking-Aloud Protocols” , it is only products
which are available, although products o f a different kind and order. N orm s are
not directly observable, then, which is all the more reason why som ething should
also be said about them in the context o f an attem pt to account for translational
behaviour.
There are tw o m ajor sources for a reconstruction o f translational norm s, textual
and extratextual:8
N O R M S IN T R A N S L A T I O N 2 07
1 textual: the translated texts themselves, for all kinds o f norm s, as well as
analytical inventories o f tran slatio n s (i.e., “ v irtu al” te xts), for variou s
prelim inary norm s;
2 extratextual: sem i-theoretical or critical form ulations, such as prescriptive
“ theories” o f translation, statements m ade by translators, editors, publishers,
and other persons involved in or connected with the activity, critical appraisals
o f individual tran slation s, or the activity of a tran slator or “ sch o o l” of
translators, and so forth.
There is a fundamental difference between these two types o f source: Texts are
prim ary products of norm -regulated behaviour, and can therefore be taken as
imm ediate representations thereof. N orm ative pronouncem ents, by contrast, are
merely ¿y-products o f the existence and activity o f norm s. Like any attem pt to
formulate a norm, they are partial and biased, and should therefore be treated with
every possible circum spection; all the m ore so since— em anating as they do from
interested parties— they are likely to lean tow ard propaganda and persuasion. There
m ay therefore be g aps, even con tradiction s, between explicit argum ents and
dem ands, on the one hand, and actual behaviour and its results, on the other, due
either to subjectivity or naïveté, or even lack o f sufficient knowledge on the part of
those who produced the form ulations. O n occasion, a deliberate desire to m islead
and deceive m ay also be involved. Even with respect to the tran slators themselves,
intentions do not necessarily concur with any declaration o f intent (which is often
put dow n post factum anyway, when the act has already been com pleted); and the
way those intentions are realized m ay well constitute a further, third category still.
Yet all these reservations— proper and serious though they m ay be— should not
lead one to abandon semi-theoretical and critical form ulations as legitimate sources
for the study o f norms. In spite o f all its faults, this type of source still has its merits,
both in itself and as a possible key to the analysis o f actual behaviour. At the same
time, if the pitfalls inherent in them are to be avoided, norm ative pronouncem ents
should never be accepted at face value. They should rather be taken as pre-
system atic and given an explication in such a way as to place them in a narrow and
precise fram ework, lending the resulting explicata the coveted system atic status.
While doing so, an attem pt should be m ade to clarify the status o f each formulation,
however slanted and biased it m ay be, and uncover the sense in which it was not
just accidental; in other words how, in the final analysis, it does reflect the cultural
constellation within which, and for w hose purposes it w as produced. A part from
sheer speculation, such an explication should involve the com parison o f various
norm ative pronouncem ents to each other, as well as their repeated confrontation
with the patterns revealed by [the results of] actual behaviour and the norm s
reconstructed from them— all this with full consideration for their contextualization.
(See a representative case in W eissbrod 1 98 9.)
It is natural, and very convenient, to comm ence one’s research into translational
behaviour by focussing on isolated norm s pertaining to well-defined behavioural
dim ensions, be they— and the coupled pairs o f replacing and replaced segm ents
representing them— established from the source text’s perspective (e.g., translational
replacem ents of source m etaphors) or from the target text’s vantage, point (e.g.,
binom ials o f near-synonyms as translational replacements). However, translation
208 GIDEON TOURY N O R M S IN T R A N S L A T I O N 2 09

is intrinsically m ulti-dimensional: the m anifold phenomena it presents are tightly

interwoven and do not allow for easy isolation, not even for m ethodical purposes.
Therefore, research should never get stuck in the blind alley o f the “ p aradigm atic”
phase which w ould at best yield lists o f “ norm em es” , or discrete norm s. Rather, it
should alw ays proceed to a “ syn tagm atic” phase, involving the integration of
normemes pertaining to various problem areas. Accordingly, the student’s task can
be characterized as an attem pt to establish what relations there are between norm s
pertaining to various dom ains by correlating his/her individual findings and
w eighing them against each other. Obviously, the thicker the network of relations
thus established, the more justified one would be in speaking in terms o f a normative
structure (cf. Jack son 19 60 :1 4 9-6 0) or model.
This having been said, it should again be noted that a tran slator’s behaviour
cannot be expected to be fully systematic. N o t only can his/her decision-m aking be
differently m otivated in different problem areas, but it can also be unevenly
distributed throughout an assignm ent within a single problem area. Consistency in
translational behaviour is thus a g rade d notion which is neither nil (i.e., total
erraticness) nor 1 (i.e., absolute regularity); its extent should emerge at the end o f a
study as one o f its conclusions, rather than being presupposed.
The A merican sociologist Jay Jack so n suggested a “ Return Potential C urve” ,
show ing the distribution o f approval/disapproval am ong the m embers o f a social
group over a range o f behaviour of a certain type as a model for the representation
disapprove
o f norms. This m odel (reproduced as Figure 1) m akes it possible to make a gradual
distinction between norms in terms o f intensity (indicated by the height o f the curve, Figure 1 S ch em atic d ia g ra m sh o w in g the R e tu rn P o te n tial M o d e l fo r
its distance from the horizontal axis), the total range o f tolerated behaviour (that representing norm s: (a) a behaviour dim ension; (b) an evaluation
part of the behavioural dim ension approved by the group), and the ratio o f one of dim ension; (c) a return potential curve, show ing the distribution of
these properties of the norm to the others. approval-disapproval am ong the m embers of a group over the whole
One convenient division that can be re-interpreted with the aid o f this model is range o f behaviour; (d) the range of tolerable or approved behaviour.
tripartite:9 Source: Jack so n 19 60.

a. B asic (primary) norm s, m ore or less m andatory for all instances o f a certain
behaviour (and hence their m inimal com m on denominator). Occupy the apex
o f the curve. M axim um intensity, m inim um latitude o f behaviour.
b. Secondary norms, or tendencies, determining favourable behaviour. M ay be
predom inant in certain parts o f the group. Therefore com m on enough, but
not m andatory, from the point o f view o f the group as a whole. O ccupy that
part o f the curve nearest its apex and therefore less intensive than the basic
norm s but covering a greater range of behaviour.
c. Tolerated (permitted) behaviour. O ccupies the rest of the “ positive” part of
the curve (i.e., that p art which lies above the horizontal axis), and therefore
o f minim al intensity.
“ A special g ro up ,” detachable from (c), seems to be o f considerable interest and
importance, at least in som e behavioural dom ains:
c’. Sym ptom atic devices. Though these devices m ay be infrequently used, their
occurrence is typical for narrow ing segments o f the group under study. On
the other h and, their absolute non -occurrence can be typical o f other
segments.
We may, then, safely assum e a distributional basis for the study o f norm s: the more
frequent a targ et-te xt ph enom en on , a shift from a (h ypothetical) adequ ate
reconstruction o f a source text, or a tran slational relation, the m ore likely it is to
reflect (in this order) a m ore perm itted (tolerated) activity, a stronger tendency, a
more basic (obligatory) norm . A second aspect o f norm s, their discrim inatory
capacity, is thus reciprocal to the first, so that the less frequent a behaviour, the
sm aller the group it m ay serve to define. At the sam e time, the group it does define
is not just any group; it is always a sub-group o f the one constituted by higher-rank
norm s. To be sure, even idiosyncrasies (which, in their extrem e, constitute groups-
of-one) often manifest them selves as personal w ays o f realizing [more] general
attitudes rather than deviations in a completely unexpected direction.10 Be that as it
may, the retrospective establishm ent of norm s is alw ays relative to the section
under study, and no autom atic upw ard projection is possible. Any attem pt to m ove
in that direction and draw generalizations would require further study, which should
be targeted tow ards that particular end.
210 GIDEON TOURY
Finally, the curve model also enables us to redefine one additional concept: the
actual degree o f conformity m anifested by different m em bers of a group to a norm
that has already been extracted from a corpus, and hence found relevant to it. This
aspect can be defined in terms o f the distance from the point o f m axim um return (in
other w ords, from the curve’s apex).
N otw ith standing the points m ade in the last few parag raph s, the argument for
the distributional aspect o f the norm s should not be pushed to o far.
As is so well know n, we are in no position to point to strict statistical m ethods
for dealing with translation al norm s, or even to supply sam pling rules for actual
research (which, because o f hum an lim itations, will alw ays be applied to samples
only). At this stage we m ust be content with our intuitions, which, being based on
knowledge and previous experience, are “ learned” ones, and use them as keys for
selecting corpuses and for hitting upon ideas. This is not to say that we should
abandon all hope for methodological improvements. On the contrary: much energy
should still be directed tow ard the crystallization of system atic research m ethods,
including statistical ones, especially if we wish to transcend the study o f norm s,
which are alw ays lim ited to one societal group at a time, and move on to the
form ulation o f general laws of translational behaviour, which would inevitably be
probabilistic in nature. To be sure, achievements o f actual studies can themselves
supply us with clues as to necessary and possible m ethodological improvements.
Besides, if we hold up research until the m ost system atic m ethods have been found,
we m ight never get any research done.
N otes
1 “ The existence o f norms is a sine qua non in instances of labelling and regulating;
w ithout a norm , all deviations are m eaningless and becom e cases o f free
variation ” (Wexler 1974:4, n. 1).
2 “ An adequate translation is a translation which realizes in the target language
the textual relationships of a source text with no breach of its own [basic]
linguistic system ” (Even-Zohar 19 75:43 ; my translation).
3 The claim that principles o f segm entation follow universal patterns is just a
figm ent of the im agination o f som e discourse and text theoreticians intent on
uncovering as m any universal principles as possible. In actual fact, there have
been various traditions (or “ m od els” ) o f segm entation, and the differences
between them always have im plications for translation, whether they are taken
10
to bear on the form ulation o f the target text or ignored. Even the segm entation
o f sacred texts such as the O ld Testam ent itself has often been tam pered with
by its tran slators, norm ally in order to bring it closer to target cultural habits,
and by so doing enhance the translation ’s acceptability.
4 Thus, for instance, in sectors where the pursuit o f adequate translation is
m arginal, it is highly probable that indirect translation w ould also become
com m on, on occasion even preferred over direct translation. By contrast, a
norm which prohibits m ediated translation is likely to be connected with a
grow ing proxim ity to the initial norm o f adequacy. Under such circum stances,
N O R M S IN T R A N S L A T I O N 211
if indirect translation is still perform ed, the fact will at least be concealed, if
not outright denied.
5 And see, in this connection, Izre’el’s “ Rationale for Translating Ancient Texts
into a M odern Lan gu age ” (1 99 4). In an attem pt to com e up with a m ethod for
translating an A kkadian myth which w ould be presented to m odern Israeli
audiences in an oral performance, he purports to combine a “ feeling-of-antiquity”
with a “ feeling-of m odernity” in a text which w ould be altogether simple and
easily comprehensible by using a host o f lexical items o f biblical H ebrew in
I s r a e li H e b re w g ra m m a tic a l an d sy n tact ic stru c tu re s . W h ereas “ the
lexicon ...w o uld serve to give an ancient flavor to the text, the gram m ar would
serve to enable m odern perception” . It m ight be added that this is a perfect
m irror im age o f the way H ebrew translators started sim ulating spoken Hebrew
in their texts: spoken lexical items were inserted in gram m atical and syntactic
structures which were m arked for belonging to the written varieties (Ben-Shahar
1 983), which also m eant “ new ” into “ o ld ” .
6 See also my discussion of “ Equivalence and Non-Equivalence as a Function of
N o rm s” (Toury 1 9 80 :63 -7 0).
7 “ There is a clear difference between an attem pt to account for som e m ajor
principles which govern a system outside the realm of time, and one which
intends to account for how a system operates both ‘in principle’ and ‘in tim e.’
Once the historical aspect is adm itted into the functional approach, several
implications must be drawn. First, it m ust be adm itted that both synchrony and
diachrony are historical, but the exclusive identification o f the latter with history
is untenable. As a result, synchrony cannot and should not be equated with
statics, since at any given m om ent, m ore than one diachronic set is operating
on the synchronic axis. Therefore, on the one hand a system consists o f both
synchrony and diachrony; on the other, each o f these separately is obviously
also a system. Secondly, if the idea o f structuredness and systemicity need no
longer be identified with homogeneity, a sem iotic system can be conceived of
as a heterogeneous, open structure. It is, therefore, very rarely a uni-system but
is, necessarily, a polysystem ” (Even-Zohar 1 99 0 :1 1 ).
8 Cf. e.g., Vodicka (19 6 4:7 4), on the possible sources for the study of literary
norm s, and W exler (1 9 7 4 :7 -9 ), on the sources for the study of prescriptive
intervention (“ purism ” ) in language.
9 Cf. e.g., H rush ov sk i’s sim ilar division (in Ben-Porat and H rushovski 1 9 74 :9-
10) and its application to the description of the norm s o f H ebrew rhyme (in
H rushovski 1971).
And see the exam ple of the seemingly idiosyncratic use o f Hebrew ki-xen as a
translational replacement of English “ well” in a period when the norm dictates
the use o f lu-vexen.