The Translation of Poetry: Some Observations and a Model

Author(s): André Lefevere

Source: Comparative Literature Studies, Vol. 12, No. 4 (Dec., 1975), pp. 384-392
Published by: Penn State University Press
Stable URL: https://www.jstor.org/stable/40246103
Accessed: 26-12-2019 19:05 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/40246103?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Penn State University Press is collaborating with JSTOR to digitize, preserve and extend
access to Comparative Literature Studies

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
 The Translation of Poetry:
Some Observations and a Model

ANDRÉ LEFEVERE

ABSTRACT

Many existing translations, versions, and imitations of poetry distort the

source text by either expanding or compressing it. This distortion is the re-
sult of the translator's exclusive concentration on one feature of the source
text (rhyme, metre, sound). This, in turn, leads to all kinds of restrictions, the
upholding of which gradually becomes as important as, if not more important
than, making the source text accessible to a new audience. Moreover, trans-
lators tend to be satisfied with a mere rendering of the linguistic elements of
the source text, ignoring those elements of it which, because they belong to a
different time, place, and tradition, must also be interpreted to the reader.
Translators should, therefore, possess the ability to understand the source text
as a total structure, to measure both its sense and communicative value and to
replace both by their equivalents in the target language, to topicalize culture-
bound and explain structure-bound elements, and to select a form within their
literary tradition which is as closely equivalent as possible to the form of the
source text. (AL)

Most current translations, versions, and imitations of poetry have

two main features in common: they both expand and compress the
source text. Compression is mainly achieved through the use of com-
pound-words or paraphrase and frequently camouflaged as "expla-
nation," "interpretation," or even "improvement." Expansion is
achieved through the use of exaggeration, explicitness, and padding,
and camouflaged in much the same way. Both procedures inevitably
result in a distortion of the target text as a literary work of art and
in a falsification of the source text.
The distortion takes place on various levels: - morphological
distortion: words in the target text are truncated, elided, or deliber-
ately used in their archaic forms. - distortion of both sense and com-
municative value:1 excessive use of archaisms, etymologisms, circum-
locutions, ready-made utterances, and tautologies. Structurally im-
portant connotations are often missed and the source text is misrepre-
sented. - distortion of syntax, either through superimposition of
source language syntactic patterns on the target language, or through
use of function words, modifiers, interjections, and unconnected
stopgaps. - distortion of structure, through the addition or omission
of structural features and the use of modulation or topicalization.
In all cases the resulting target texts singularly fail to qualify as liter-
ary translations, as translations which can both exist as a literary
work of art in their own right and give the reader an accurate impres-
sion of what the source text is like.
Phonemic translation (the attempt to render mainly the sounds of
the source text in the target text) works best when it translates least.

384

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
 TRANSLATION OF POETRY 385

It is moderately successful only in its ono

its caiques of proper names. It results in the
syntactic jigsaw puzzles and is faintly access
who are already familiar with the source tex
practical point of view, it is useless. If one u
to put sound first, one is forced to use an e
and obsolete words and, in doing so, one con
stant use of the dictionary. If, however, the
language dictionary all the time to understa
reading, he might just as well use a source-l
dictionary and discard the translation altoge
hand, the reader is satisfied with a reproduc
source text only, there is no reason at all wh
source text aloud without understanding a s
result will not be very different from the o
of a phonemic translation.
In the wider context of the evolution and
literatures, phonemic translation is positivel
trating on sound only, it distorts all the oth
text and reduces it to a curiosum, a bilingua
survival in the literature of the target langu
lator clearly has little or no desire to make
to his readers; he merely wants to experime
rather, he is consciously trying to dictate th
language. By forcing the sound patterns of
he purports to "bridge the gap" between l
any attempt at bridging gaps or breaking do
otherwise, may be, the attempt to "fuse" lan
the start because it is an attempt made by o
whereas language itself is the product of a w
Literal translation is a myth because "no tw
cal, either in the meaning given to correspo
ways in which such signals are arranged in p
stands to reason that there can be no absolu
tween languages. Hence there can be no fully
Literal translation pays an enormous price f
elusive "accuracy." It strips the source text o
ary characteristics and, in doing so, "stérilis
des classiques étrangers, dont elle dégoûte
translation has, moreover, all too often been
lation as such, so that what is true in itself,
be a fully exact translation, has increasingly
against all translation, and more especially a
literature. It is therefore absolutely necessar
idea of a "literal translation" for the myth
prove that translation of literature, of poet

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
386 COMPARATIVE LITERATURE STUDIES

once it has been freed from the obligation to render

sense. It follows that, as I hope to show later, the tra
poetry is mainly concerned with matching communic
communicative value.
Are literal translations entirely without justification? They will
probably mean something "to the student of anthropology, history
and semantics, but to the student of literature they are meaningless."4
They are also harmful, especially in the teaching of literature, since
they force the teacher all too often to fall back on pseudo-literary
texts mummified into monstres sacrés with no appeal at all to the
student, who reads them only because the heavy hand of fate has
put them on the syllabus.
There is a line of defence behind which advocates of literal trans-
lation frequently retreat. Let us briefly consider their argument.
Literal translations, it runs, are of course not literature, nor were
they ever meant to be. They are intended only to serve as a help to
the reader who is unable to struggle through the source text on his
own, but who has enough knowledge of the source language to plod
through the source text if he is allowed the occasional glance at the
literal translation often printed opposite. However, the many "im-
provements," archaisms, and ready-made utterances encountered in
literal translations suggest that the literal translation is definitely
intended to be some form of literature, an antiquated one, no doubt,
but the result does not invalidate the translator's intention. This is
precisely what makes literal translations unacceptable from the liter-
ary point of view. Not only does it strip away the genuine literary
characteristics of the source text, it also insists on replacing them
with pseudo -literary elements in the target text. Moreover, the many
interpretations and explanations smuggled into the text, some of
which are inevitable precisely because "the semantic map of each
language is different from those of all other languages"5 but the
majority of which are traceable to the translator's whim, make any
literal translation's claim to accuracy highly doubtful. The only
way in which a really literal translation could be of any help to the
half -bilingual reader is in the form of an interlinear version, not as
a hybrid creation forever vegetating on the boundary between the
literary and the non-literary.
Literal translation is often defended on grounds slightly beyond
the level of rational discourse. One is the familiar "bridging the gap
between languages" argument, this time through the superimposition
of syntactic source language patterns on the target language. Another
ground could be called the "elitist" attitude. Literal translation is,
indeed, often advocated by teachers of either the classics or "exotic"
languages. They feel that their wisdom should definitely not be shared
with the masses. On the contrary, the masses will have to make the
effort to read everything in the original language. In the meantime,

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
 TRANSLATION OF POETRY 387

any translation will do. This elitist attitu

age in which the student of literature in
for access to classical texts for the under
tures, and also the need for access to the
The ideal student of literature would, acc
tion, have to know Hebrew, Sanskrit, Gr
European languages, Chinese, a major In
To be able to master all these languages w
lifetimes, let alone their literatures. Tran
second best, but that is no reason to with
dent of literature the best possible equiv
an essential literary text.
There are also those who advocate literal translation on national-
istic grounds. No foreign literature can, according to them, be of any
value for the evolution of their native literature. Consequently, any
foreign text should be condescendingly treated as a curiosum which
should, logically, be made to sound as "curious" as possible in trans-
lation. Both the literal and the phonemic translation will qualify for
this unenviable role. The literal translation has the unquestionable
advantage that it sells better without, at the same time, becoming a
threat to the self-sufficient parochialism of the native literature.
Both metrical translations and translations of poetry into prose
are caught in a vicious circle. By focusing his attention exclusively
on one external feature of the source text, the metrical translator
merely destroys the balanced structure of that source text, destroy-
ing it, in fact, as a work of literature. He is left with a curiosum, a
catalogue of verbal and metrical acrobatics. If, on the other hand, he
wants to maintain the structure of the source text, he has to abandon
his attempt at primarily rendering its metre in the target text. The
prose translator is faced with a similar dilemma. If he wants to make
good the losses in communicative value a word incurs in the course
of its transfer from poetry into prose, he is forced either to distort
the equivalent word in the translation or drag in modifiers, circum-
locutions, metaphorical expressions, and explanatory phrases: a
mass of additional words which inevitably weighs down the syntax
of the target text. If, however, he wants to retain the rhythm of the
source text by either superimposing its syntactic patterns on the
target text or constructing long rhythmical periods, he inevitably
distracts from the communicative value of the single word.
Translators of poetry into rhyming and blank verse have the
technical skill needed to render the source text competently into
a rhyming or blank verse target text, but little more. In the rhyming
translation, rhyme and metre reveal themselves as tyrannical forces
and many important features of the original have to be sacrificed to
their demands. The rhyming translator fights a losing battle against
self-imposed restrictions. The target text reads more often than not

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
388 COMPARATIVE LITERATURE STUDIES

like an unintentional parody of its source. If the blan

lator wants to adhere to his metrical scheme by eithe
contracting his lines, or to break away from it by ei
ing a different pattern or by highlighting a word or
is forced to resort to a number of contortions and verbal bravura
pieces which make it very difficult for him to produce a target text
completely free from clumsiness and verbosity.
In short, the reason why so many current translations, versions,
and imitations are unsatisfactory renderings of the source text is
simply this: they all concentrate exclusively on one aspect of that
source text only, rather than on its totality. This exclusive concen-
tration can be traced back to either idiosyncratic or traditional
motivations. The motivation is obviously idiosyncratic in the case
of the phonemic translator, the writer of versions, and the imitator.
It results from an excessive respect for tradition in the case of the
metrical, the rhyming, the blank verse, and the prose translator.
Once this unfortunate starting point is established, the process by
which the actual translations come into being is much the same in
all cases. The one aspect of the source text he sets out to "render
faithfully" soon restricts the translator's freedom of action in the
most stringent way for any translation. Literary translation starts
with an attempt to render as faithfully as possible the source text's
communicative value. The implications are that words from various
stylistic spheres in the source text must be rendered by words from
matching stylistic spheres in the target text. On no account must
the result be allowed to amount to a weakening of the communi-
cative value of the target text as compared to that of the source text.
This weakening is caused by either a failure to grasp the communi-
cative value of words in either source or target language, or in both,
or by the policy of subordinating the communicative value of words
to the imperatives of sound, metre, prose, or rhyme.
The same operation must be carried out in the fields of time,
place, and tradition. The only difference is that careful consider-
ation should now be given to the structural value of time, place, and
tradition elements in the over-all framework of the source text. This
is probably the most delicate problem with which the translator of
a literary text is confronted. Sense and communicative value can,
on the whole, be measured with relative ease and accuracy against
the "common usage" prevalent in source and target language at
the time of the writing of the source and target text.
But how is one to measure the value of time-place-tradition
elements contained in the source text? If one consistently modern-
izes, "presents in such a language as would be used by the authors
themselves if they lived in our days and wrote using our language,"6
one is bound to end up with ludicrous anachronisms. If, on the
other hand, one consistently refuses to modernize, one leaves the

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
 TRANSLATION OF POETRY 389

reader with inexplicable passages. The crit

be the structural value or the time-place-
source text. Consequently, the translator
between what belongs, in the source text
the time (what is culture-bound) and wha
right understanding of the source text as
bound). Structure-bound elements should
text without modernization; they should,
tained. They must also, as economically
explained within the target text. It is also
those time-place-tradition elements in th
the reader's mind, be easily linked to ana
place, and tradition of the target text. M
obligatory in all other cases if the transla
his task, even if this means losing some o
the source text. In this case, translation p
"second best" only, but the loss in subtlet
allowed to occur after the most careful d
be more than offset by the correspondin
the other hand, the translator refuses to
within the target text, the reader will eit
forced to make continuous use of the lex
hardly be able to react to the target text
literary art. Alternatively, the communic
reader will be constantly interrupted by
notes, or else certain caiques in the text w
"exoticism."
There is another culture-bound element which helps to shape the
whole structure of the source text: the literary tradition in which
that source text originated. Here again this element of the "code"
circumscribing the "theme" in the source language's literary tradition
should be replaced by an element of the "code" which occupies
roughly the same position in the literary tradition of the target lan-
guage, provided that the use of this element of the code does not ex-
cessively narrow down the translator's field of action, that it does
not, in other words, become a self-imposed restriction.
It is not my aim to compile an exhaustive list of "rules," careful
observance of which will automatically lead translators to the ideal
translation. Since the literature of the target text is constantly chang-
ing and, if one translates contemporary authors, the literature of the
source language as well, to say nothing about the interrelationship
between the two, no list of rules could ever claim to be exhaustive.
It would, in fact, be a thing of the pedantic past as soon as it was
formulated. What I have in mind could therefore more aptly be
described as a tentative inventory of the competence the literary
translator should possess in order to perform satisfactorily when

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
390 COMPARATIVE LITERATURE STUDIES

translating different source texts belonging to diff

and traditions. This competence consists of four fac
important.
(1) The ability to comprehend the source text as a whole, as a total
structure, rather than the "negative capability" of concentrating on a
single aspect of the source text, and consequently the ability to realize
that time-place-tradition elements contained in the source text should
receive the same attention, be transposed in the same way and with the
same care as the linguistic elements. This of course presupposes an ex-
pert knowledge of the source text's literary, social, and cultural back-
ground by the translator. Mere linguistic knowledge is insufficient, and
"poetic inspiration" will hardly make up for the absence of sound
scholarship. The translator will therefore have to combine "Sprach-
kenntnis mit kùnstlerischer Veranlagung. . . . Der blosse Philologe
gleicht bei der Ubersetzungsarbeit einem sezierenden Arzt, der einen
Organismus in seine kleinsten Teile zerlegt, und ihn dann nicht mehr
zusammensetzen kann: er hat die Teile in der Hand, fehlt leider nur
das geistige Band. Der Dichter ohne Sprachkenntnisse bleibt zu
dilettantischer Stumperei verdammt."7
(2) The ability to measure the communicative value as well as the
sense of the source text, and consequently the ability to replace it by
a target text which approximates, as closely as possible, the same com-
municative value. This implies that all distortions (morphological,
semantic, syntactic, structural) are warranted only by corresponding
distortions in the source text. Conversely, it also implies that any
conscious distortions in the source text should be rendered in the
target text, and not smoothed over or "silently corrected." It fol-
lows that the translator will have to have not only the stylistic ability
to judge the source text adequately but also the stylistic facility to
recreate it in the target text. Anybody with the relevant linguistic
training can more or less satisfactorily perform the first task. Very
few are able to successfully perform the second.
(3) The ability to distinguish between culture -bound and structure-
bound time-place-tradition elements in the source text and conse-
quently the ability to modernize the former and to retain the latter
while explaining them within the target text, with the proviso that
no modernization is needed in the case of time-place-tradition ele-
ments either explained by their context or easily connected, in the
reader's mind, with analogous elements in the time, place, and tra-
dition of which he is a part. A translation can only be said to be
complete if and when both the communicative value and the time-
place-tradition elements of the source text have been replaced by
their nearest possible equivalents in the target text. This will often
imply changes, from slight to radical, in both structure and texture
of the source text, but they will be changes inspired by the desire
to render the source text accessible and will therefore be motivated

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
 TRANSLATION OF POETRY 39 1

by an underlying faithfulness to its t

attention given to only one of its aspe
the source text to the taste of both tr
nor, finally, by mere whim (imitation
when, and only when, the source text
new audience. It does not imply pander
new audience's taste.
(4) The ability to select, within the literary tradition of the target
language, a form which will most closely match the position the source
text occupies in the literary tradition of the source language, with,
however, the all-important proviso that the chosen form will not be
allowed to restrict the translator's range in selecting communicative
value equivalents, nor that it will force him to produce a target text
which would be considered "unacceptable" (in the linguistic sense)
by the majority of the speakers of the target language. If no such
form is available, then it is the translator's privilege and obligation
to create one, just as it is his privilege to create new communicative
values to match values contained in the source text. In doing so, the
translator allows both the foreign literature to influence and further
the development of his native literature and the foreign language to
enrich his own language. This implies that the translator must not
only know the language but also the literary tradition of the source
text and, no less important, that stylistic facility in the target lan-
guage has to be supplemented by constant contact with the literary
evolution taking place in that target language. On another level, this
implies that, in the process of translation, source language has to
encounter target language on a footing of equality. Schleiermacher's
notorious maxim, "Entweder der Ubersetzer lâsst den Schriftsteller
môglichst in Ruhe, und bewegt den Leser ihm entgegen, oder er
lâsst den Leser môglichst in Ruhe und bewegt den Schriftsteller
ihm entgegen,"8 holds true only if for some reason either the lan-
guage of the author or that of the reader is looked upon as in some
sense superior. It is not true when source and target language meet
on a basis of equality in the person who combines the functions of
both author and reader: the translator.
To the four factors already enumerated, a fifth one could be added,
which helps explain why it is Utopian to expect that "rules" will
automatically lead any translator to achieve a satisfactory result:
the ability to interpret the theme of the source text in the same way
as the original author. This is the dividing line between translation
proper and version/imitation. The version writer basically shares the
original author's interpretation of the theme, but not enough to
respect the way in which the original author expressed it in the
code that determines the actual shape of the source text. He does,
therefore, only render the content, while altering the form. He
does, in other words, not substitute an equivalent code, but a code of

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms
392 COMPARATIVE LITERATURE STUDIES

his own, on the assumption that this code will ma

more easily accessible. The version, therefore, nev
impression of the source text the way the transla
basically an exercise in re-writing, not an attempt
thor's real interpretation of the theme accessible t
Imitation has very little to do with translation as
writes a different work, using the source text on
inspiration.
The translator, as opposed to the version writer
should possess the ability to re-interpret the sour
lines of the interpretation laid down by the origin
should not superimpose his own interpretation on
word, try to achieve an "equivalent effect:" "that
best which comes nearest to creating in its audien
sion as was made by the original on its contempor
lator should neither wilfully heighten (version) or
(translations concentrating exclusively on one aspe
text) that effect.
No translation will, of course, completely achiev
wirklichung der im Original recht vernommenen
Mitteln der eigenen Zunge."10 The "ideal" transl
which is based on neither term of the antiquated o
fidelity and freedom but rather on the "principle
the original on the overall level, combined with a
treatment of details,"11 might, however, be consi
cessful than most current types of translation of

ANDRÉ LEFEVERE • University of Ant

NOTES

l.On "sense" and "communicative value," see G. Leech, Semantics (Harmondsworth,

1974),Ch.ii. _ _ _
2.E.A. Nida, Towards a Science of Translating (Leiden, iyb*j, p. 100.
3. G. Mounin, Les Belles Infidèles (Fans, iy&5), p. 84.
4.P.T. Manchester, "Verse Translation as an Interpretive Art, Hispania, AAAiv
(1951),p.68. _
5.J. Green
b.Mathesius quoted in V. Frochazka, "Notes on iransiatmg lecnmque in r.i*
Garvin, ed. A Prague School Reader (Washington, 1964), p. 95.
7.H. Gipper, Sprachliche und Geistige Metamorphosen bet Gedichtubersetzungen
(Dusseldorf,1966),p.69. .... _ . __ mm
8. Quoted in H.J. Storig, éd., Das Problem des Ubers
9.E.V. Rieu, "Translation" in Cassels Encyclopedia
p. 555. ___. _
10. W. Schadewaldt, He
11. A. ropovic, ine Lioncept aniri or repression in îrausutiiuu n.iutiy»i» mj^.
Holmes, éd., The Nature of Translation (Den Haag, 1972), p. 80.

This content downloaded from 154.59.124.74 on Thu, 26 Dec 2019 19:05:41 UTC
All use subject to https://about.jstor.org/terms