Translating poetic songs

An attempt at a functional account of strategies

Peter Low
University of Canterbury, New Zealand

Poems have often been turned into songs, notably as German Lieder.
Classical singers use translations of these in several different ways: as cribs
for themselves, in printed programmes for their audiences, as singable ver-
sions, etc. Since no single target-text is ideal for all of these purposes, the
Skopostheorie of Hans J. Vermeer may help translators to match their strate-
gy with the particular skopos (“goal or purpose”) of their translation. The
author identifies five specific functions which a song-translation may serve,
and proposes a range of five translation strategies intended to match these
particular skopoi. A demonstration is given of how these strategies produce
different English versions of a few lines from a Baudelaire poem.

Keywords: art-song, Lieder, poem-translating, skopos, translation theory

Introduction

This article concerns the translating of those poetic texts which have come to be
used as “Art Songs”. Some well-known examples are the poems of Goethe
which Schubert turned into Lieder and the Verlaine poems set to music by
Debussy: these have been presented many times as songs to audiences of
different languages. The texts’ authors intended them to stand alone as poems
in their own right, and published them as words without music. Hence they can
be distinguished from song-lyrics, which are texts written by song-writers or
lyricists for the express purpose of being set to music. Although this distinction
is not always a clear one — there exist gray areas and cross-overs — it can be
made easily in most cases, particularly in European works after 1800. A song-
lyric is born into the song genre; a poem, born as poetry, may later “acquire

Target 15:1 (2003), 91–110.

issn 0924–1884 / e-issn 1569–9986
© John Benjamins Publishing Company
92 Peter Low

dual nationality” by being used as a song-text, in a different part of that

language’s cultural polysystem, and perhaps even in a different period (as for
example when sonnets by John Donne were set to music by Benjamin Britten).1
What happens then when art songs are presented to an audience unfamiliar
with their language? Their verbal content may be lost, so that they are appreciat-
ed by the listeners only as abstract music, albeit prompted by a text. Such
appreciation will depend on the musical idiom, which may indeed be familiar
within this other cultural system, as (say) Mendelssohn’s musical style was well-
known in Victorian England. Frequently, however, attempts are made to
transfer the songs’ verbal content into some form accessible to the target
audience — attempts at translation. Such attempts should ideally consider not
only the words of the song-texts but also the wider literary and sociocultural
aspects of what Even-Zohar (1990) has called inter-systemic transfer.
Because the texts in question are genuine creative works of “high” literary
art, this topic belongs with translation of literature, and is related to the long-
standing debates about how to translate poetry (e.g. Weissbort 1989, Holmes
1988). However, the questions I am posing here are ones that are seldom
covered in discussions of literary translation: they are questions of purpose and
function. Even with poetry translation, not all the important questions concern
the source text: some of them concern the needs and expectations of the people
for whom the target text is intended. This is not altogether a new thought — the
main reason why translators used to turn unrhymed Latin odes into rhyming
French or English ones is that the target cultures expected poems to rhyme (e.g.
Marsh 1941).
Most discussions of poetry-translating say little about the purpose and
function of a translation. For example, André Lefevere’s Translating Poetry
focuses on the difficult problem of “making the ST available as a literary work
of art in the target language” (1975: 37). He gives some attention to contextual
issues of time, place and tradition; yet he does not mention the possibility that
a poem may have a second role as a song-text, or that the TT may have a very
different cultural purpose from the ST. Older writers about poem-translation seem
to make several implicit assumptions. They assume that since the ST is a written
text, usually printed in a book or magazine, the resulting TT will be presented in
the same form. As for function and purpose, they also seem to assume either that
the TT will attempt to fulfil for its target audience all the functions which the ST
fulfilled for its original target audience — or simply that its purpose will be for
individual reading (in circumstances where the reader can readily pause to
reflect, and may sometimes be able to consult reference books).
 Translating poetic songs 93

Yet these matters are far from self-evident. For instance, much ancient and
traditional poetry was solely oral. Even today, many poems have oral-auditory
features; some are recited at poetry readings; and a few poets choose to call
themselves “performance poets”. It follows, therefore, that just as drama-
translators should consider how well their TTs will work in live theatrical
performance, so poetry-translators should ask themselves how much impor-
tance they should give to oral-auditory features in their TTs, and to what extent
they wish their versions to be recited. To take a famous example: Alexander
Pope’s rhyming Iliad shows that he expected oral presentation, in the salons of
18th-century England. More recently, however, E. V. Rieu’s English prose
version of the Odyssey showed little interest in phonic effects, and nor did
Schadewaldt’s German version, which was meant to be read like a novel. Even
Burton Raffel, whose study The art of translating poetry devotes a chapter to
translating oral poetry, can discuss a written poem like Baudelaire’s “Le Balcon”
(1988: 86–90) without mentioning that a translation might be made primarily
for actors wishing to recite it, or for singers wishing to perform it in Debussy’s
musical setting.
A further reason for asking questions about purpose is the well-known fact
that translated poems cannot replicate all the features of the originals, and often
do not even come close. As a genre, poetry is notable for the interactions it
creates between form and content, the attention it gives to phonic effects, the
importance it places on connotational meanings, its frequent use of metaphor,
its condensed style, and its uses of non-standard word-order. In poetry transla-
tion, as David Connolly puts it : “it is impossible to account for all the factors
involved and to convey all the features of the original in a language and form
acceptable to the TL culture and tradition” (Baker 1998: 171). In view of this
impossibility, one should logically expect that some focus on function and
purpose would help a translator to decide which features to prioritise in a given
case and which may be sacrificed at less cost.
In translation theory, various scholars such as Katharina Reiss and Hans J.
Vermeer have emphasised the importance of purpose and the needs of the
end-user of the TT. Vermeer uses the term skopos to designate the “goal or
purpose, defined by the commission and if necessary adjusted by the transla-
tor” (2000: 230). Although he pays more attention to informative texts, he tries
to use skopos thinking as a basis for a general translation theory, and therefore
considers it applicable to expressive texts as well. Christiane Nord, another
“functionalist”, has written specifically about functionalism in literary translation
(1997: 80–103). The present paper suggests that skopos may have particular
94 Peter Low

relevance to the situations that arise where poetic texts are set to music, noting
these theorists’ view that if the skopos requires a change in function between ST
and TT, “the standard will not be intertextual coherence but adequacy or
appropriateness with regard to the skopos” (Nord 1997: 33; Reiss and Vermeer
1984: 139).
Questions of function and purpose are certainly complex when we consider
the various uses made of poetic texts in European Classical Music. This is a
large field, where a subculture of music-lovers in (say) the English cultural
polysystem can enjoy the artistic products of other languages conveyed by the
non-verbal vehicle of a musical tradition which crosses national and linguistic
boundaries. In vocal music words are often prominent, albeit generally subordi-
nate to the musical considerations. It is common in an evening’s concert or
recital for a singer to perform in two or more languages, of which the local
audience might know only one. There may be Lieder in German, mélodies in
French, other songs in Russian or Spanish, arias in Italian, sacred works in
Latin… Some recordings are similarly polyglot. Given this situation, it is
essential that the skill of translators be deployed, both for the sake of the perform-
ers (so that they can render the words well) and for the sake of the listeners (to give
them at least some idea of the verbal dimension of the performance).
The present article focuses on the case of “Art Songs”, where the difficulties
of translating are particularly great, because here the words receive greater
respect than is common in other types of song (such as operatic arias), and
because the source texts often have a high aesthetic value as poems. Indeed
composers often chose texts for their aesthetic value (for example Debussy or
Reynaldo Hahn, whose literary tastes were highly developed). And composers
often took special steps to ensure that the words are clearly audible — quiet
accompaniment by piano alone, medium tessitura (because high-pitched
phrases make words hard to catch), little use of melismas (which stretch
syllables over several notes), and special focus on particular words.
Furthermore, some composers sought to match the specific mood of a given
verbal phrase by appropriate musical techniques, be they melodic, harmonic or
dynamic. Consequently, there are some songs where the music is puzzling or
unconvincing unless you understand the words. A good example is Debussy’s
“Colloque sentimental”, where the words (by Verlaine) are paramount, and
where there are musical discontinuities resulting from the changing voices and
emotions in the text. But even in cases where the music by itself has proved to
be appealing — as in “Danse macabre”, by Saint-Saëns, of which the composer
later made purely instrumental versions — full appreciation of the song
 Translating poetic songs 95

certainly includes a grasp of the text (an outrageous text, in this case, by Henri
Cazalis). For those art songs which most depend for their effect on verbal
comprehension, good translations are particularly important — and they are
not easy to do.
For the translator, however, there is one consolation: a TT used in the field
of classical music is not a “stand-alone” text. Usually the composer has already
“interpreted” it by non-verbal means, by musical devices intended to convey
something of the ST’s beauty or emotional power. The song-translator is
therefore in a slightly different position from normal poetry-translators (who
are sometimes accused of replacing good poems with a bad ones). Here — even
when the song is not sung in the original — the composer shares with the
translator the difficult task of communicating to the TL audience the subtlety
and richness of the poet’s text. Good performers help too.
Writing about text function, Christiane Nord has suggested that “the target
text should be composed in such a way that it fulfils functions in the target
situation that are compatible with the sender’s intention” (1997: 92). Although
“sender” there denotes the text’s author, we might suggest in the case of songs
that the composer is a secondary sender, whose intentions should (preferably)
be compatible with the author’s, and that the target situation in the concert-hall
is created largely by the singer, who is the “user” of this verbal-musical message.
Let us examine more closely these target situations, and the function of the TTs
within them.

A diversity of functions

The translations used in the field of Art Song appear in various forms, for
various purposes. Most fall into one of five categories:
a. A performer’s crib
Editions of vocal music commonly print translations for the singer to use as a
crib while rehearsing the song. These TTs may appear near the start of the
volume, or at the start of each song, or in the end-papers. Often such transla-
tions are presented in parallel format; sometimes the layout is interlinear.
b. A recording insert
Recordings of classical vocal music almost always print translations as well as
the original words. This enables the listener to read a selected version while
96 Peter Low

listening to the recording, or to study it at another time, comparing it with the

original (or with the other versions). Some CD inserts print parallel texts in two
languages — or even four.
c. A programme text
Frequently the printed programme for a concert gives translations of the foreign
texts into the main language of the audience. The expectation is that many
people will consult this before or during the performance, and may retain it for
later reading. Sometimes recital programmes print the ST also, in a parallel or
“facing-page” format, for the sake of those who have some familiarity with the
source language.
d. A spoken text
Sometimes a singer or presenter reads aloud a TT before the musical number is
performed in the foreign language. This may be a full version of a short poem,
or a précis of a longer one. Some radio stations broadcasting classical songs use
brief introductions (in their own language) giving a gist of the verbal content.
e. A sung text
Sometimes a “singable version” of a song is prepared, published — and even
sung. Also called “singable versions” these are TTs designed to fit the music and
to be performed with it, so that the SL (source language) is not heard at all.
Even that list of five is not quite exhaustive. Two others are:
f. surtitles, which are occasionally screened in concert halls, as is now com-
mon in opera-houses; and
g. subtitles, which are attached to videos of song-performances. These two
forms of TT differ from all the above, involving distinct constraints of space and
time. They impose their own specific translation strategies, which will not be
explored in this paper.
The above outline of purposes and functions compels me to conclude that no
single translation would ever suit all these purposes equally well. For example, a TT
devised to be read in a concert programme (c) will prove useless to a singer actually
wishing to sing the song in that target language — the rhythms will certainly be
wrong. But a TT devised to be a “sung text” (e) will not suit function (a) — having
given careful attention to matching the music, it will be too inaccurate structurally
and semantically to assist the singer in untangling the ST. These points may be
obvious to scholars, but they have not always entered the mind of a singer or
concert-agent who requests “the translation” of a particular song-text.
 Translating poetic songs 97

A mistake that can be observed too often, even in sophisticated English

cities, is for a “performer’s crib” version (function a) to be printed by musicians
in a recital programme (function c), where easy comprehensibility is needed by
the audience in the concert-hall. Such TTs, even when they achieve semantic
accuracy, tend to be too stilted and unnatural for this skopos. Necessarily, they
focus on “the trees not the forest” — they translate at the level of words and
phrases rather than sentences, stanzas or larger meaning units, precisely because
their validity lies in helping the singer to untangle the SL grammar. To use them
in a recital programme is unfair both to the poets (who never wrote in trans-
lationese) and to the composers (who chose the texts on aesthetic grounds). It
is also discourteous to the people in the concert-hall, imposing on them
unnecessarily the irksome task of deciphering awkward chunks of the TL.
But surely a single translation can fulfil more than one function? Yes, some
can. Of the five functions listed above, the two most similar are the programme
text (c) and the recording insert (b). For straightforward poems lacking
complexities of meaning or quirks of style, a single TT will serve both purposes
rather well. For complex poems, however, a translator may wish to reduce the
processing effort imposed on concert-goers, and may judge that they need a
more easily digestible version than that served to record-lovers who are listen-
ing at home, who can pause their stereos between songs, and for whom the
musical performance has no visual dimension.
The above pages have attempted to clarify the functional questions which
concern song-translating. Before proposing my own answers, let me consider
the thoughts of some more experienced practitioners.

Some practitioners’ thoughts

As it happens, scholars have written very little about the strategic questions of
song-translating, as distinct from poetry-translation in general. The Routledge
encyclopedia of Translation Studies (Baker 1998) has no article or index-entry on
the subject. And a commendable article by Arthur Graham (1989) considers
only one purpose or function ((e), discussed below). In search of wisdom and
clarity, I turn therefore to the words of some experienced practitioners who
have written short descriptions, though not theoretical analyses, of their
approaches to translating song-texts in English.
The earliest page I wish to discuss comes from The ring of words: An
anthology of song texts. Here the translator, Philip Miller, says:
98 Peter Low

Generally the English versions are given line by line, and as often as feasible the
word order has been retained, so that the meaning of the original words may
be made reasonably clear. Occasionally it has seemed permissible to read
between the lines — some of the German poetry is notably obscure. … In such
cases, and in others where strict literalness results in a sort of pidgin English, I
have preferred to keep my translation readable. (1963: xxviii)

In advocating readability, Miller shows a somewhat functional approach. But his

defensive tone suggests that he expected his readers to desire “strict literalness” —
something which he knew translators cannot always deliver. The phrase “it has
seemed permissible to read between the lines” is strangely apologetic. Since
these texts are lyric poems, and since a competent translator must first be a
competent reader, it is not only permissible but actually essential for a translator to
read between the lines, and to grasp the sub-text, what various scholars call the
“implicature” of the ST (Baker 1992:223). The translator, having grasped it, has an
opportunity to make that implicature more explicit in the TT — this may often be
desirable, particularly if the alternative would be a TT more obscure than the ST.
Perhaps the largest compendium of song-translations is The Fischer-Dieskau
book of Lieder (1977). This is a “reference book for the recital-goer, record-lover
and musician”. It offers nearly 400 pages of parallel-text translations, and
devotes a generous half-page to a “Translators’ Note”. The translators in
question, George Bird and Richard Stokes, say, among other things:
We have tried to convey the full sense, and as much as possible of the tone of
the originals. To this end, we chart a middle course between, on the one hand,
the extreme of literal prose translation in which all poetic semblance is stifled,
and on the other, the extreme of verse translation in which sense and the order
of lines are distorted for the sake of the metre and rhyme … So far as possible,
we observe the German line order. (Fischer-Dieskau 1977: 7)

This note spells out some general intentions — the wish to help readers to
engage with the original text (observing line order), the wish to render tone as
well as sense — and an explicit decision not to replicate phonic features of the
ST, at least not slavishly.2 It rejects two strategies as inappropriate: verse
translation and unpoetic literal prose. And it continues thus:
We have resisted the temptation to explain, either in footnotes or in the
process of translation, lines which depend for their sense on facts not explicit
in the poem. (Fischer-Dieskau 1977: 7)

As an example of this, Bird and Stokes cite Goethe’s line “So lasst mich schein-
en, bis ich werde”, which is the start of a poem set by Schubert, Schumann, and
 Translating poetic songs 99

Wolf. They decided not to elucidate this line, despite knowing, from its original
context in Wilhelm Meisters Lehrjahre, that it means “Let me remain attired as
an angel”. Their TT — “So let me seem, until I am” — must surely seem
opaque to many readers, some of whom might misconstrue this English “so” as
meaning “therefore” rather than “thus”.
In my view, this is a poor option resulting from a narrow or blinkered
conception of what constitutes a text. The word “Engel” can be detected in the
sub-text of this line, because Goethe undoubtedly placed it there. It is part of
the implicature of the text, and therefore the translators were wrong to ignore
it — though they may argue that subsequent lines give the reader a good chance
of guessing. If they had written: “Let me appear like an angel, until I am one”,
thus explaining the meaning “in the process of translating” (as they put it), then
they would have been conveying relevant information which few readers
already possessed. I note that more recently, Stokes opted to provide occasional
footnotes in a subsequent work, mostly to elucidate proper names in some
French song-texts (Stokes 1999: xviii.).3
A feature of the above-mentioned works is their strategic decision to
proceed line by line. This decision seems intended to meet the particular needs
of singers who are ignorant of the SL, and is justified for that purpose. For other
purposes, however, a line-by-line approach is unduly restrictive: it means that
some of the normal tools of competent translators — tools such as transposi-
tion, modulation, paraphrase, compensation in place — are rendered unavail-
able, totally or partially. Given this constraint, and the difficulties found in
some of the STs, the translators’ work is performed skillfully. Undoubtedly,
numerous compromises are involved; sometimes they are in awkward English,
and sometimes even “line-by-line” is still not sufficiently “word-for-word” to
enlighten an ignorant singer about the meaning of every SL word.
A less compromising approach is taken by one of the best-selling compen-
diums of song-translations, Lieder line by line, and word for word by Lois
Phillips, which first appeared in 1979. Unlike the works just cited, Lieder line by
line adopts a double strategy of providing both a pedantically close version of
the ST (in interlinear format) and a “clear prose version” intended “to disentan-
gle the often unintelligible series of words resulting from a literal word-for-
word translation” (Phillips 1979: vi). The former version is designed to help
people to engage with the German words and word-order; the latter gives a
better sense of the whole poem. More recently, Timothy Le Van (1991) has
taken a similar dual-text approach.
100 Peter Low

For a long time, of course, the “word-for-word” approach to translation has

been criticised by scholars. They have almost always been right: for most
purposes, this approach is at best a “a pre-translation process” (Newmark
1988: 11). The worst examples of this approach look like the work of novices
with poor dictionaries and untenable views about “literal translation” produc-
ing maximum accuracy.
In the hands of Phillips, however, such an approach produces texts which
are functional: they appropriately match a particular skopos. While her word-
for-word TTs are awkward, much more so than those of the translators men-
tioned above, they are sufficiently clear when studied carefully. And she makes
no apology for their awkwardness (although she shies away from the extreme of
rendering “es gibt” as “it gives”). The point is that these word-for-word versions
are not expected to stand alone: thus she happily offers lines such as “Saw a boy
a (little) rose to stand” — Goethe!? — knowing that readers will compare this with
the prose version alongside, which says: “A boy saw a wild rose growing”. To me,
a word-for-word translation is never adequate for poetic texts, and is acceptable
only when combined with another version that reads more like poetry.
One might question whether this translator’s “clear prose versions” are the
perfect complement: the willingness of English-language translators to render
foreign poems into prose certainly neglects part of their true character. But
there is no doubt that her dual formula is functional: together, the two TTs are
very helpful to singers and other anglophones with little sense of German. They
manage to provide views of “both the trees and the forest”.
This provision of two different TTs shows that Lois Phillips has made the
conscious assumption (a valid one) that not all singers have the same foreign-
language skills. We might add that the reading of translated poems itself
requires significant language skills, particularly if the translations are clumsy
ones. Her word-for-word versions remain, in my opinion, of little use to readers
with absolutely no German. But the intended target reader is someone with
enough sense of SL grammar and vocabulary to be truly enlightened by the
versions offered. It is worth noting that the training of professional singers
nowadays places more weight on language skills than it did formerly (in the
days when many songs and operas were performed in translation).
Besides, by providing a second version of each ST, Lois Phillips shows aware-
ness of the shortcomings of her “word-for-word” TTs. The “clear prose versions”
which she offers, though intended principally to assist students, would be quite
appropriate also as spoken texts — purpose (d) above — if the song is short.
 Translating poetic songs 101

Cultural and contextual issues

The relatively brief “translators’ notes” cited above do not tackle the broader
cultural issues involved in translating poetic texts for readers in a different time
and place. Some of these issues concern factual knowledge about specific items
mentioned in the ST; some concern more subtle allusions to cultural practices.
The general problem is that some poems can be fully appreciated only by people
who possess that contextual knowledge. By and large, translators of songs have
been silent on these problems — they seem to have thought that their brief was
only to render the overt semantic content of the poems, despite the fact that poetry
is the kind of literature most interested in allusive and connotational meanings.
Considerations of purpose and function, however, ought to lead us to a
different view. Because the role of translators is part of an intercultural commu-
nication process, their task must be not just to look back at the ST but also to
“look forward” at the needs of the TL reader who will use the TT within the
target culture. In the case of song-texts, the typical reader of translations is often
a music-lover rather than a poetry-buff, and cannot be assumed to be the
“serious reader of poems” which discussions of poetry-translation tend to have
in mind. This less sophisticated reader may be ill-equipped for teasing meaning
out of an opaque text, and may thus require a more helpful translator. Here is
a case where considerations of skopos should affect the translators’ handling of
cultural and contextual details. The differences in cultural knowledge between
the ST addressees and the TT audience may, as Nord puts it, “require an
adjustment of the relationship between explicit and implicit information in the
text” (1997: 63).
A few examples can help to pin-point these issues:
i. Translators of Brahms’s well-known song “Sapphische Ode” (words by
Hans Schmidt beginning “Rosen brach ich”) have always simply called it
“Sapphic Ode”, never showing any concern that this might mislead modern
readers into expecting Lesbian content in the text.
ii. People keen to understand Ernest Chausson’s song “Les heures”, which has
words by Camille Mauclair, may well be unaware that the title denotes
ancient female divinities; and will probably be misled by translations for
line 9 (“et certains, blêmes”) which fail to indicate that these adjectives,
being masculine, do not refer to the Hours.
iii. A Verlaine poem from the 1890s, “L’incrédule” (set by Reynaldo Hahn)
begins: “Tu crois au marc de café,/Aux présages, aux grands jeux…”. Given
the cultural context, and the poem’s contrast between a sceptical speaker
102 Peter Low

and a superstitious addressee, these “grands jeux” can be identified as Tarot

cards. A translator who does not convey this fact is omitting relevant
information.
iv. When Apollinaire called a poem “A la Santé”, and when Arthur Honegger
set it to music, neither saw any need for a footnote explaining that “La
Santé” is a prison. Yet an explanation would assist readers unfamiliar with
Paris, who might otherwise guess it was a sanatorium.
In raising these matters, the present article may seem to be requesting that the
translator become an elucidator or annotator of texts, like the contributors to
Burnshaw’s volume The poem itself (Burnshaw 1960). I do not wish to go too far
in this direction: I think that good translators need real understanding of the
ST, but not that all their knowledge of culture-specific details needs to be
inserted into the TT. In cases where relevant cultural knowledge was possessed
by the original readership and not by the TL readership, I claim that there is a
strong case for providing it. This is common practice already in informative
texts. Although a few readers may feel insulted by such glosses, they should see
that others will benefit.
Judgement must be exercised, of course. And one of the criteria to consider
is skopos. When a translator decides to elucidate a text, the specific function of
the TT can help to determine which choice to make out of several available
options. For example “sur le Sèvres” (in Mallarmé’s poem “Placet futile”, set by
Debussy and Ravel) cannot be satisfactorily rendered as “on the Sèvres”. The
translator may choose from several options: “on Sèvres porcelain”, “in a rococo
painting on porcelain”, or “on a piece of Sèvres*” with footnote. But to provide
an opaque TT, with important mysteries unsolved, is discourteous to readers.
The risk of opacity is compounded by the style of those 20th-century
writers who use language in idiosyncratic, even deviant ways. For example,
many songs by Francis Poulenc have texts by Apollinaire or Eluard, poets who
abandoned the use of punctuation, expecting the syntax and rhythms of their
lines to rule out most ambiguity. Translators who do not insert commas and
stops, however, are likely to make their TTs even more opaque than the STs, for
example leaving greater doubt about where a sentence begins or ends. I consider
that the judicious insertion of punctuation is needed to improve functionality.4
The above reflections lead me to suggest a range of translating strategies,
each of which tries to take account of the specificities of a particular purpose. As
Vermeer puts it, the skopos can “help to determine whether the source text needs
to be ‘translated’, ‘paraphrased’, or completely ‘re-edited’” (Vermeer, 231).
 Translating poetic songs 103

Proposed strategies

a. For a performer’s crib: Gloss translation

This kind of TT is to be studied, not simply read once. In my view, a “per-
former’s crib” calls for a particularly broad view of the content of a ST. Many
poems carry meanings located “between the lines” — allusive or connotational
meanings. It is up to translators to enlighten singers about these “sub-textual”
matters, at least when the impact of a ST depends heavily on them. For exam-
ple, when a French song mentions “la Durance”, both poet and composer will
have assumed that everyone knows what it is. The translator, however, may opt
to spell out the most relevant information with or without the use of square
brackets — “the Durance [river]” — or else place it in a footnote: “a river in
south-east France”. There may even be a place for cultural glosses, definitional
translations, and “split translations” offering dual versions of a word or phrase.
One term that matches this kind of TT is “gloss translation”. As used by
Eugene Nida, this term means a translation “designed to permit the reader to
identify himself as fully as possible with the person in the source-language
context”. Gloss translations seek “formal equivalence” rather than what Nida
calls “dynamic equivalence”. The translator “attempts to reproduce as literally
and meaningfully as possible the form and content of the original”, and even
uses footnotes “in order to make the text fully comprehensible” (1964: 159).
For the other purposes (below), the options of footnotes or editorial
brackets are either impossible or undesirable.
b. For a recording insert: Semantic translation
A recording, by definition, can be used in different times and places. Few
generalisations can be made about the target audience. With songs in (say)
English, even native speakers appreciate the chance to read the ST, since the
singers’ lips and gestures cannot be watched.
For translators, the presence of the printed ST makes some difference. They
may judge that “parallel text” format reduces their freedom to rearrange the
line-structure of the original.
In practice, TTs in recording inserts often choose to observe the line-order
of the ST. This makes the pages available for use by the “semi-bilingual” reader:
someone with enough sense of SL grammar and vocabulary to be truly enlight-
ened by the versions offered.
The kind of translating which I propose for this purpose is a sophisticated
one acknowledging that the text is a piece of creative writing in the SL culture,
104 Peter Low

and helping the readers to understand why it won the composer’s attention. It
is not intended for study, however, and is not as free to expand and explicate as
a gloss translation may. The best existing term I have found for this is Peter
Newmark’s term “semantic translation”. This is generally written at the
linguistic level of the SL author, though it is not inflexible and “may make small
concessions to the readership”. It takes some account of the aesthetics of the ST;
it allows for “the translator’s intuitive empathy with the original” (1988: 46–47).
c. For a programme text: Communicative translation
A text for use in a recital or concert programme needs to be reader-friendly, so
that it can be digested in a relatively limited time. For the kind of translation
most suitable, I use Newmark’s word “communicative” translation, a kind
which he distinguishes from “semantic” in various subtle ways. In his terminol-
ogy, a communicative TT is written at the linguistic level of the intended
readers, and attempts “to render the exact contextual meaning of the original in
such a way that both content and language are readily acceptable and compre-
hensible to the readership” (Newmark 1988: 46–47.).5
Now Newmark’s own recommendation is that communicative translation
be used for texts of any informative type, but not for expressive texts like poems
where literary style has a high importance (and can be better respected by
semantic translation). Thinking more functionally, however, I consider that the
circumstances of the concert-room mean that priority should go to the needs of
the readers. It is possible to make some intelligent generalisations about the
average audience’s knowledge of cultural and linguistic details. The translator
can assess, for example, which minor details in the ST may be omitted at little
cost, which cultural details may be “domesticated” into the TL culture, and
which proper names may require some elucidation in the TT.
Such TTs may be printed in prose format, or as unrhymed free verse. When
a parallel text is not going to be printed, translators are free to view a whole
sentence or stanza as the “unit to be translated”. They should, however,
maintain the larger structures of the ST as formulated by the poet and subse-
quently set to music by the composer.
d. For a spoken text: Gist translation
A TT which a singer will read aloud before singing the song needs to be short
and digestible. Unless the ST is very brief, a précis is needed, a simplified “gist
translation” that will render the main sense and feeling of the words without
too much detail.
 Translating poetic songs 105

In my view, it is best for singers themselves to devise such texts. This is

because even singers with limited knowledge of the SL should be encouraged to
wrestle with the words, as part of their overall preparation for performing a
song. Usually it would be easier for a translator to devise such a version; but
there are real artistic benefits to be gained from imposing this task on the
singer, who may choose to consult a native speaker or a printed translation
(such as a performer’s crib version).
e. For a sung text: A singable translation
Singable versions are not currently fashionable in classical music. This cannot
be attributed solely to snobbery. The two main reasons are the strong claim of
the ST, and the defective nature of most TTs. Only the ST offers the actual
words set by the composer, with all their phonic features such as rhymes and
vowel-sounds, and of course their integral meaning.6
As for singable TTs, although people have often attempted to devise them,
it is a difficult task. Translators are subject to huge constraints imposed by the
pre-existing music, because they cannot ignore the rhythms, the note-values,
the phrasings or the stresses of the music — even phrasings and pitch-levels
may have to be considered. Those working into English must also seek to
reduce the number of short vowel-sounds and the clustering of consonants.
Ideally, the TT must sound as if the music had been fitted to it, even though it
was actually composed to fit the ST. It is not surprising, then, that many of
these TTs are too badly done to be usable — many are so marred by forced
rhymes and unnatural language that performers simply cannot sing them with
conviction. As Arthur Graham (himself a singer) has put it: “Most translations
don’t ring true as texts a sensible composer might have set” … “The singer
needs words that may be sung with sincerity” (Graham 1989: 31, 35).
Yet Graham considers — and so do I — that talented translators applying
well-judged strategies can produce acceptable results. The argument in favour
of singable translations points out that most songs were intended to communi-
cate with audiences verbally as well as musically. This argument has its greatest
persuasive strength when applied to the kinds of songs, notably comic or
dramatic ones, which stand to gain most from being comprehended instantly
and directly by their audience. One such song which I have attempted to
translate is Emmanuel Chabrier’s “Villanelle des petits canards”, where the
quirky music attempts to mimic the waddling ducklings. A mere written
translation of Rosemonde Gérard’s cute poem — no matter how lively — will
not elicit the full humour which Chabrier’s song conveys to a French audience.
106 Peter Low

The devising of this kind of TT presents the translator with fascinating

challenges, on which the present article will not expand. Such singable versions
are more common in popular commercial music than in the classical field,
chiefly because the target audience of popular songs is different and has
different expectations.

A comparative example

To illustrate more clearly the differences in detail which result when transla-
tions are devised for different purposes, I have concocted a situation where one
ST has generated five different TTs. In the English versions given here (my
own), the differences are visible even with purposes (b) and (c), which are the
two most similar. The ST chosen is the start of a sonnet by Charles Baudelaire,
“La vie antérieure” (1857), which was magnificently set to music by Henri
Duparc (1884). Its first quatrain reads:
J’ai longtemps habité sous de vastes portiques
Que les soleils marins teignaient de mille feux,
Et que leurs grands piliers, droits et majestueux
Rendaient pareils, le soir, aux grottes basaltiques.

a. performer’s crib
For a long time I lived under vast porticoes
which the suns of the sea coloured with a thousand fires;
and the great pillars [of these porticoes], straight and majestic,
made them look similar, at evening, to caves of basalt.
 Translating poetic songs 107

Note that this TT is best not read aloud, because most listeners would interpret
the plural “suns” as the homophone “sons”.
b. recording insert
For a long time I lived under vast porticoes
which the ocean sunshine coloured with a thousand fires;
their great columns, which were straight and majestic,
made them look similar, in the evenings, to basalt caves.

If this were printed in a CD insert using parallel format, the presence of the ST,
printed alongside, might incline some translators to opt for English words
cognate with the French ST (e.g. “grottoes” rather than “caves”). My own
inclinations tend towards non-cognate words, since I prefer in general to
translate common French words with frequently used English ones.
c. programme text
For a long time I lived by the sea in a palace
which was lit up in fiery colours by the ocean sunshine.
Its vast porticoes had tall, majestic pillars,
which, in the evenings, made them look like basalt caves.

This TT, unlike (a) and (b), seeks easy digestibility. The shifting of the phrase
“vast porticoes” from line 1 to line 3 would not be attempted if the ST were
printed alongside.
d. spoken text
In a previous life I lived by the ocean in a tall palace from which I could see
tropical sunsets reflected in the surging waves.

The length of the complete ST (14 lines) means that a summary is needed. This
TT is a paraphrase condensing not four but eight lines, and mentioning the title
as well. It is given only as an indication of what a “gist” approach might produce.
e. sung text
For a long time I dwelt under porticoed halls
which ocean sunshine tinged with light of many flames
and whose majestic pillars standing straight and tall
made them appear, at dusk, like vast palatial basalt caves.
108 Peter Low

On the page, this TT looks clumsy. Its rhythm, however, is based not on the
prosody of the French poet (or of any English ones), but on that specific music
— which varies from two to six syllables per measure and seldom places the
longest notes at the downbeats. The sole purpose of such a TT is to make the
song available for singing in the target language. Other translators have at-
tempted the different task of making the original poem available as a literary
work of art in English. James McGowan, for example, turns for this purpose to
the traditional English iambic pentameter:
I once lived under vast and columned vaults
Tinged with a thousand fires by ocean suns,
So that their grand, straight pillars would become,
In evening light, like grottoes of basalt. (McGowan 1993: 31)

Given that “La vie antérieure” is a fine poem by a great poet, I expect to be told
that my versions fail to do justice to the ST. My reply is that they are to be
judged not solely with retrospective reference to the French poem, but first and
foremost prospectively: how appropriate are they for their intended function?
This question — the key question in skopos theory — is not the only good
question. I submit, however, that it is a valid question that deserves to be placed
alongside two broader questions, which even functionalist thinking recognises
as important. These are the literary question: “How well is Baudelaire respected
in these TTs?” and the verbal-musical question: “How best can Duparc’s fine
song be made available to English-speaking audiences, not just as a melodic
vocalise ending with a strange postlude on the piano, but as a genuine song
marrying music to words?”7
 Translating poetic songs 109

Notes

1. Some song-lyrics are more poetic than others. The most text-focused of jazz songs, for
example, pose problems similar to those treated here.
2. Much the same thinking is repeated by one of these men twenty years later in A French
song companion: “As far as possible”, says Stokes, “I have observed in translation the French
line-order, and have attempted to render the sense and tone of the poem without any slavish
adherence to the original rhymes and metre” (1999: xviii).
3. Stokes is very sparing with these footnotes, however, allegedly “fearful of dumbing down”
— though I doubt that footnotes could reduce anyone’s knowledge or intelligence.
4. Pierre Bernac’s 1970 book on French Art Song effectively adds punctuation to these poets’
original texts by indicating — for singers — where to breathe and where to observe liaisons.
There are no punctuation marks, however, in the accompanying English translations, made
by Winifred Radford, who calls them “literal line-by-line” versions (Bernac 1970: xiv).
5. An earlier work by Newmark devotes many pages to communicative and semantic
translation, with illustrative examples (Newmark 1982).
6. This is true even when a composer uses texts that themselves are translations — for
example when Schubert set German versions of Shakespeare lyrics.
7. The present writer often translates songs. Sometimes these are cribs for performers,
sometimes they are singable versions. The most numerous have been devised for concert
programmes. Dozens of these can be seen on the web at a site called “The Lied and song texts
page” — http://www.lieder.net/ Hundreds of translations are viewable there, which vary in
strategy, purpose and quality. Those which I have provided are mostly intended for
programmes and recordings (purposes (b) and (c)). They seek to use an English which is
reasonably intelligible and idiomatic — as the original French poem usually was — and
which shows some of the poetic qualities that encouraged the composer to choose the ST.

References

Baker, Mona. 1992. In other words. London and New York: Routledge, 1992.
Baker, Mona, ed. 1998. Routledge encyclopedia of Translation Studies. London and New York:
Routledge.
Bernac, Pierre. 1970. The interpretation of French song. London: Gollancz.
Burnshaw, Stanley, ed. 1960. The poem itself. New York: Holt, Rinehart.
Even-Zohar, Itamar. 1990. “Polysystem studies”, special issue of Poetics today 11:1.
Fischer-Dieskau, Dietrich. 1977. The Fischer-Dieskau book of Lieder. New York: Knopf.
Graham, Arthur. 1989. “A new look at recital song translation”. Translation review 29. 31–37.
Holmes, James S. 1988. Translated!: Papers on literary translation and translation Studies.
Amsterdam: Rodopi.
Le Van, Timothy. 1991. Masters of the French Art Song. Metuchen NJ: Scarecrow.
110 Peter Low

Lefevere, André. 1975. Translating poetry: Seven strategies and a blueprint. Assen: Van
Gorcum.
Marsh, Edward, trans. 1941. The Odes of Horace. London: MacMillan
McGowan, James, trans. 1993. The Flowers of Evil. Oxford: Oxford University Press.
Miller, Philip. 1963.The ring of words: An anthology of song texts. New York: Norton.
Newmark, Peter. 1982. Approaches to translation. Oxford: Pergamon.
Newmark, Peter. 1988. A textbook of translation. New York: Prentice Hall.
Nida, Eugene A. 1964. Toward a science of translating, with special reference to principles and
procedures involved in Bible translating. Leiden: Brill.
Nord, Christiane. 1997. Translating as a purposeful activity. Manchester: St Jerome.
Phillips, Lois. 1979. Lieder line by line, and word for word. London: Duckworth.
Raffel, Burton. 1988. The art of translating poetry. Pennsylvania and London: Pennsylvania
State University.
Reiss, Katharina and Hans J. Vermeer. 1984. Grundlegung einer allgemeinen Translations-
theorie. Tübingen: Niemeyer.
Stokes, Richard and Graham Johnson. 1999. A French song companion. Oxford: O. U. P.
Vermeer, Hans J. 1989. “Skopos and commission in translational action”. Lawrence Venuti,
ed. The Translation Studies reader. London and New York: Routledge, 2000: 221–232.
Weissbort, Daniel, ed. 1989. Translating poetry. Basingstoke: MacMillan.

Résumé

On a souvent transformé des poèmes en chansons, notamment dans le Lied allemand. Les
chanteurs classiques se servent de traductions à diverses fins: comme un aide-déchiffrage
pour eux-mêmes, sous la forme d’un texte imprimé pour leurs auditeurs, en tant que version
“chantable”, etc. Puisqu’une seule traduction ne peut assumer toutes ces fonctions, la
“théorie du Skopos” de Hans J. Vermeer pourrait aider les traducteurs à adapter leur stratégie
au skopos (but poursuivi) en question. L’auteur identifie cinq fonctions possibles d’une
chanson traduite, et propose, corrélativement, une gamme de cinq stratégies, chacune
convenant à un skopos spécifique. Ensuite, il applique ces stratégies à des transpositions en
anglais de quelques vers de Baudelaire.

Author’s address
Peter Low
Department of French
University of Canterbury
CHRISTCHURCH
New Zealand
e-mail: [email protected]