The Oxford Handbook of Translation Studies

Kirsten Malmkjær (ed.), Kevin Windle (ed.)

https://doi.org/10.1093/oxfordhb/9780199239306.001.0001
Published: 2011 Online ISBN: 9780191744020 Print ISBN: 9780199239306

CHAPTER

12 The Translation of Poetry 

Francis R. Jones

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

https://doi.org/10.1093/oxfordhb/9780199239306.013.0013 Pages 169–182
Published: 18 September 2012

Abstract
Poetry translation may be de ned as relaying poetry into another language. Poetry's features can be
sound-based, syntactic or structural or pragmatic in nature. Apart from transforming text, poetry
translation also involves cognition, discourse, and action by and between human and textual actors in
a physical and social setting. A poetry translation project usually aims to publicize a poet or poets.
Poetry translation is typically overt. Poetry translators are concerned to interpret a source poem's
layers of meaning, to relay this interpretation reliably, and/or to ‘create a poem in the target language
which is readable and enjoyable as an independent, literary text. Poetry translation involves challenges
and these are highlighted in this article. Poetry accounts for a tiny proportion of world translation
output. Case studies and examples taken from poetry, however, have dominated theory-building in
translation studies at the expense of more frequently translated genres.

Keywords: poetry translation, cognition, discourse, poet, interpretation, translation output, theory
building
Subject: Translation and Interpretation, Linguistics
Series: Oxford Handbooks

Introduction

Poetry
Poetry translation may be de ned as relaying poetry into another language. Poetry is regarded here as a
genre of literary text, and genre as a socially de ned cluster of communication acts. These have rules that
are largely pre-agreed by communicators (poets, publishers, audiences, say), though they may also be
negotiated on the spot (Andrews 1991: 18; Stockwell 2002: 33–4). Some of poetry's rules might specify its
typical textual features. For examples, let us look at Yù jiē yuàn (‘Jade stairs lament’) by Tang dynasty
Chinese poet Li Po, with modern pronunciation and Ezra Pound's 1915 English version added (Preminger,
Brogan, and Terry 1993, Matterson and Jones 2000; texts from Bradbury n.d., Pound 1949):
 graphic yù jiē shēng bái lù The jewelled steps are already quite white with dew

graphic yè ju qīn lúo wà It is so late that the dew soaks my gauze stockings

graphic què xià shŭi-jīng lián And I let down the crystal curtain

graphic líng-lóng wàng qīu yuè And watch the moon through the clear autumn

Some of poetry's features are sound-based, such as line-length (here, ve syllables) or onomatopoeia (líng-
lóng, meaning ‘jade-tinkling’ or ‘exquisite’, sounds like tinkling jade). Some are syntactic or structural,
such as the parallel verb—adjective—noun syntax and high-rise—fall tones of shēng bái lù (literally ‘grows
p. 170 white dew’) and qīn lúo wà (‘invades net stockings’). Others are more pragmatic in nature, such as

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

ambiguity and multiple meaning (does líng-lóng here mean ‘jade-tinkling’, ‘exquisite’, or both?), or image
and metaphor (e.g. shŭi-jīng lián, ‘quartz-crystal blind’, also refers to the tears of the concubine waiting all
night in vain for the emperor). Poetry may deviate from prose norms of syntax or collocation, as with the
highly compressed syntax of Tang poetry. Moreover, poems often combine many of these features in a
restricted space, making them potentially the ‘most complex of all linguistic structures’, with a ‘special
relationship between form and meaning’ (Holmes 1988: 9, Boase-Beier 2009). Some see poetry's
communicative e ect as made up of more than denotative meaning: the last two lines, for example, allude
to the sound of crystal-bead blinds in the autumn wind, the passing of youth, and more besides. Linked to
this is a communicative purpose that is emotive or spiritual, say, rather than just informative or
transactional.

No one of these aspects, however, is enough to de ne a message as a poem, and each may also occur in other
genres (literary prose, say, or advertisements), though the more aspects it has, the more ‘poetic’ it is likely
to seem. In practice, however, communicators usually agree quickly which genre is operating. Important
here are ‘metatextual’ features whose main role is to de ne genre, such as framing signals announcing the
genre (e.g. the word Poems on a book cover), or a special tone of voice when speaking or graphic layout when
writing.

Translating poetry
Seeing genre as communication implies that poetry translation involves not only transforming text but also
cognition, discourse, and action by and between human and textual actors in a physical and social setting
(Buzelin 2004; 2005: 736–40; Jones 2009). Pound's translation act above, therefore, involves textual
changes (jade to jewelled, say). Cognitive factors may underlie this change: as Pound read no Chinese, for
example, he might not have regarded jade's connotations of ‘precious, royal’ as self-evident.
Interactionally, Mary, widow of Asia scholar Ernest Fenollosa, was a key human actor: she gave Pound
access to Ernest's draft translations (textual actors) because she saw links between Pound's imagism and
Ernest's ideas on Chinese poetics (Wilson 2004). And socioculturally, Pound's free-verse renditions were
in uential for the adoption of free verse as the default form in twentieth-century US poetry.

12.1 Introduction

12.1.1 Poetry
Poetry translation may be de ned as relaying poetry into another language. Poetry is regarded here as a
genre of literary text, and genre as a socially de ned cluster of communication acts. These have rules that
are largely pre-agreed by communicators (poets, publishers, audiences, say), though they may also be
negotiated on the spot (Andrews 1991: 18; Stockwell 2002: 33–4). Some of poetry's rules might specify its
typical textual features. For examples, let us look at Yù jiē yuàn (‘Jade stairs lament’) by Tang dynasty
Chinese poet Li Po, with modern pronunciation and Ezra Pound's 1915 English version added (Preminger,
Brogan, and Terry 1993, Matterson and Jones 2000; texts from Bradbury n.d., Pound 1949):
 yù jiē shēng bái lù The jewelled steps are already quite white with dew

yè ju qīn lúo wà It is so late that the dew soaks my gauze stockings

què xià shŭi-jīng lián And I let down the crystal curtain

líng-lóng wàng qīu yuè And watch the moon through the clear autumn

Some of poetry's features are sound-based, such as line-length (here, ve syllables) or onomatopoeia (líng-
lóng, meaning ‘jade-tinkling’ or ‘exquisite’, sounds like tinkling jade). Some are syntactic or structural,
such as the parallel verb—adjective—noun syntax and high-rise—fall tones of shēng bái lù (literally ‘grows
p. 170 white dew’) and qīn lúo wà (‘invades net stockings’). Others are more pragmatic in nature, such as

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

ambiguity and multiple meaning (does líng-lóng here mean ‘jade-tinkling’, ‘exquisite’, or both?), or image
and metaphor (e.g. shŭi-jīng lián, ‘quartz-crystal blind’, also refers to the tears of the concubine waiting all
night in vain for the emperor). Poetry may deviate from prose norms of syntax or collocation, as with the
highly compressed syntax of Tang poetry. Moreover, poems often combine many of these features in a
restricted space, making them potentially the ‘most complex of all linguistic structures’, with a ‘special
relationship between form and meaning’ (Holmes 1988: 9, Boase-Beier 2009). Some see poetry's
communicative e ect as made up of more than denotative meaning: the last two lines, for example, allude
to the sound of crystal-bead blinds in the autumn wind, the passing of youth, and more besides. Linked to
this is a communicative purpose that is emotive or spiritual, say, rather than just informative or
transactional.

No one of these aspects, however, is enough to de ne a message as a poem, and each may also occur in other
genres (literary prose, say, or advertisements), though the more aspects it has, the more ‘poetic’ it is likely
to seem. In practice, however, communicators usually agree quickly which genre is operating. Important
here are ‘metatextual’ features whose main role is to de ne genre, such as framing signals announcing the
genre (e.g. the word Poems on a book cover), or a special tone of voice when speaking or graphic layout when
writing.

12.1.2 Translating poetry

Seeing genre as communication implies that poetry translation involves not only transforming text but also
cognition, discourse, and action by and between human and textual actors in a physical and social setting
(Buzelin 2004; 2005: 736–40; Jones 2009). Pound's translation act above, therefore, involves textual
changes (jade to jewelled, say). Cognitive factors may underlie this change: as Pound read no Chinese, for
example, he might not have regarded jade's connotations of ‘precious, royal’ as self-evident.
Interactionally, Mary, widow of Asia scholar Ernest Fenollosa, was a key human actor: she gave Pound
access to Ernest's draft translations (textual actors) because she saw links between Pound's imagism and
Ernest's ideas on Chinese poetics (Wilson 2004). And socioculturally, Pound's free-verse renditions were
in uential for the adoption of free verse as the default form in twentieth-century US poetry.
 Translation projects

This chapter focuses on poetry translation projects which serve real-world audiences—visually or audially,
p. 171 via print publications, websites, or streaming audio/ video, live broadcasts, public readings, etc. A project
usually aims to produce a ‘text complex’ containing more than one poem—an on-line poetry website, a
Chinese poets' session at a Dutch poetry festival, etc. Translated poems may be part of a multi-genre and/or
multi-language complex (a French literary journal combining modern poetry and prose from France and
Korea, say), or may form the complex's main element (such as a Hungarian edition of Shakespeare's
sonnets). The latter may be a target-language edition of a pre-existing complex (as with the Hungarian
Shakespeare), or may be specially assembled. If specially assembled, it may be a selection from one poet's
oeuvre, which typically aims to show the ‘best re ections of an individual poet's genius and speci city’
(Bishop 2000: 61). Or it may be a multi-poet anthology, which may well establish a poetry canon within the

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

receptor culture (Barnaby 2002: 86). Poems maybe presented bilingually, in both source and target
versions: if printed, the purpose may be to recognize that the source and target text give di erent reading
experiences. Conversely, if poems are published monolingually (in the target language only) this may re ect
the publisher's or editor's feeling that translations should not be judged against their source (Peter Jay,
Anvil Press: personal communication), though Bishop feels that it risks licensing versions that are adapted
towards receptor-culture norms (2000: 62).

A project usually aims to publicize a poet or poets. In countries which publish little translated poetry,
projects typically introduce new poets via an established publisher's ‘brand’ (Sampson 2001: 82). Projects
may also aim to construct or validate an image of the wider source culture in the receptor culture or
internationally (Lefevere 1975: 106–7). Some projects also promote a certain point of view about the source
culture, which may have a political or ideological dimension: for example, Agee (1998) presents Bosnia as a
modern European nation rather than a hotbed of warring nationalisms.

12.2 Translation projects

This chapter focuses on poetry translation projects which serve real-world audiences—visually or audially,
p. 171 via print publications, websites, or streaming audio/ video, live broadcasts, public readings, etc. A project
usually aims to produce a ‘text complex’ containing more than one poem—an on-line poetry website, a
Chinese poets' session at a Dutch poetry festival, etc. Translated poems may be part of a multi-genre and/or
multi-language complex (a French literary journal combining modern poetry and prose from France and
Korea, say), or may form the complex's main element (such as a Hungarian edition of Shakespeare's
sonnets). The latter may be a target-language edition of a pre-existing complex (as with the Hungarian
Shakespeare), or may be specially assembled. If specially assembled, it may be a selection from one poet's
oeuvre, which typically aims to show the ‘best re ections of an individual poet's genius and speci city’
(Bishop 2000: 61). Or it may be a multi-poet anthology, which may well establish a poetry canon within the
receptor culture (Barnaby 2002: 86). Poems maybe presented bilingually, in both source and target
versions: if printed, the purpose may be to recognize that the source and target text give di erent reading
experiences. Conversely, if poems are published monolingually (in the target language only) this may re ect
the publisher's or editor's feeling that translations should not be judged against their source (Peter Jay,
Anvil Press: personal communication), though Bishop feels that it risks licensing versions that are adapted
towards receptor-culture norms (2000: 62).

A project usually aims to publicize a poet or poets. In countries which publish little translated poetry,
projects typically introduce new poets via an established publisher's ‘brand’ (Sampson 2001: 82). Projects
may also aim to construct or validate an image of the wider source culture in the receptor culture or
internationally (Lefevere 1975: 106–7). Some projects also promote a certain point of view about the source
culture, which may have a political or ideological dimension: for example, Agee (1998) presents Bosnia as a
modern European nation rather than a hotbed of warring nationalisms.
 Transforming poetic text

The main task of poetry translators is to translate. I rst examine this as what might be termed ‘cognitive
habitus’ (a cluster of socially de ned information-processing practices), then in terms of cognitive
challenges and processes, and end by discussing a ective (emotional) factors.

Cognitive habitus
Here we look at how poetry translating processes are conceptualized within poetry production, translation,
p. 172 and consumption ‘ elds’, or loose-knit networks of users, texts and institutions (Inghilleri 2005b: 134–
5)—for example, poetry translators within a certain country, or receptor-language poets. As socially
mediated concepts vary across time and place, I should point out that my claims may be biased towards

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

recent European practices.

Poetry translation is typically overt. Target readers know they are reading a translator's interpretation of a
source-language poem (Boase-Beier 2004: 25–6). Hence translators may be less free than original poets to
ignore their readers' needs and abilities, and readers may read translated poems more critically than non-
translated poems.

Poetry translators are typically concerned to interpret a source poem's layers of meaning, to relay this
interpretation reliably, and/or to ‘create a poem in the target language which is readable and enjoyable in its
own right, with merit as an independent, literary text’ (Phillips 2001: 23–4; cf. Boase-Beier 2004: 25–6;
Lefevere 1975; Honig 1985: 177; Flynn 2004: 281–2; Jones 2006a). This triple habitus, of course, guides
overt translation in all literary genres. In poetry, however, surface semantics and underlying imagery are
often so closely and complexly bound with linguistic form that it is notoriously di cult to interpret these
relationships and reproduce them in a foreign-language text that meets Phillips's quality demands. This
has inspired a popular discourse of poetry translation as loss. Some lament the loss of source-text
reproduction, as in Lefevere's view that most poetry translations ‘are unsatisfactory renderings of the
source text’ because they fail to capture its totality (1975: 99). Others lament the loss of target-text quality,
as in Robert Frost's reputed saying that ‘poetry is what is lost in translation’ (in Untermeyer 1964: 18). In
practice, translators face a range of output options (Boase-Beier 2009, Hanson 1992). Prose translations
convey only source semantics, usually to help readers read the source (receptor-language poets in
collaborative partnerships, say: see below). What might be called re-creative translations aim to convey not
just the source's complex link between imagery and ‘core sense of the words’ (Reynolds 2003: 108) but also
its poetic e ects, in a viable receptor-culture poem. Rather than trying (and inevitably failing) to nd exact
‘equivalents’ for all a source poem's features, re-creative translators seek ‘counterparts’ and ‘analogues’
(Holmes 1988: 53–4). Self-reports and source—target text comparisons indicate that most translators
advocate this approach, even those who see themselves primarily as receptor-language poets (see e.g.
Honig 1985; Hughes 1989: 17–18). Some receptor-language poets, however, prefer writing ‘adaptations’
(Mahon 2006, Paterson's ‘versions’, 2006): poems more loosely based on other-language sources.

Re-creative translating is potentially the most challenging approach, for it demands three expertises of the
translator: expert poetry-reading ability in the source language; expert poetry-writing ability in the
receptor language; and mediating between the demands of ST loyalty and TT quality (cf. Keeley 2000: 19).

p. 173 Challenges
Here we look more closely at some of the challenges which re-creative poetry translators face, and the
solutions they may choose—whether for deliberate poetic e ect or because they see no better alternative.

First, reading a source poem can involve recognizing and interpreting a highly complex set of meanings and
poetic features. These may even be intentionally obscure—with modernist verse, for example (Bouchard
1993: 149).

According to Boase-Beier, when translating it is crucial to stay true to a source poem's style (its ‘perceived
distinctive manner of expression’: Wales, cited in Boase-Beier 2006: 4), because style encodes the source
writer's attitude towards the content (2004: 28–9). Stylistic loyalty is rarely straightforward, however, as
the following paragraphs show.
 With poetic form, Holmes sees translators as choosing between three main approaches (1988: 25–7):

• Mimetic: replicating the original form. This implies openness to the source culture's foreignness
(Holmes 1988: 25–6). However, the form may carry di erent weight in the receptor culture (Hejinian
1998, Ra el 1988)—a ve-syllable line feels ‘classical’ in Chinese, for example, but may seem radically
compressed in French.

• Analogical: using a target form with a similar cultural function to the source form (e.g. the English
iambic pentameter for the Chinese ve-syllable line). This implies a belief that receptor-culture
poetics has universal value (Holmes 1988: 26).

• Organic: choosing a form that best suits the translator's ‘own authenticity’ of response to the source
(Scott 1997: 35). This stresses the impossibility of recreating the source form—content link (Holmes

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

1988: 28).

According to Holmes (1988: 54), nding close correspondence between source and target is easier with ‘a
poem that leans very close to prose’, but the more complex the poem, the bigger the compromises that
translators have to make. This may mean that sub-genres of poetry di er in di culty: narrative verse, say,
may give the translator more room for manoeuvre than the more compressed lyric verse (Davis 1996: 31–2).

Some translators, particularly into languages with a strong free-verse tradition, advocate free-verse
translations of xed-form (rhyme and/or rhythm-based) source poems, often on analogical or organic
grounds. Others argue that this risks losing crucial stylistic e ects. Thus, regarding Hughes's free-verse
renderings of Pilinszky's Hungarian verse, Csokits writes: ‘without the softening e ect of the original metre
and rhyme scheme […] they sound harsher and Pilinszky's view of the world appears grimmer’ (1989: 11).

p. 174 Similarly, attitudes towards ST rhyme range from abandonment to re-creation (users 1998), with partial
preservation (replacing full by half-rhyme, say) as a compromise. Key arguments for abandonment are:

• Rhyme may have negative associations (e.g. old-fashioned or trite) for receptor-genre readers.

• Finding rhyme-words is di cult, especially when the receptor language has less exible word order
and/or a greater variety of word-endings than the source (e.g. English relative to German and Italian
respectively: Osers 1996; Feldman 1997: 5).

• Seeking rhyme leads to unacceptable semantic shifts—such as having ‘to add images that destroy the
poem's integrity’ (Bly 1983: 44–5).

Those who advocate recreating a source poem's rhyme scheme, whether mimetically or analogically, admit
that this requires technical skill, but argue that rhyme is an integral part of the poem's meaning: ‘if one
disapproves of rhyme in poetry, one should not translate poems that rhyme’ (Barnstone 1984:50–51; cf.
Mo ett 1989, 1999). Moreover, though seeking rhymes may give radical shifts in surface wording, the
underlying images can be preserved (Jones 2007).

Source poets may deliberately use ‘marked’ language varieties: language that, relative to the standard
variety, is distinctively archaic or modern, informal or formal, regional, speci c to poetry or typical of other
genres, or simply idiosyncratic. Alternatively, language varieties that might have seemed unmarked to the
poet may appear non-standard to most modern readers. Translators then face a choice between:

• replicating the source variety. This may not always replicate its e ect, however: archaisms, for
example, may seem original and exciting to modern Serbo-Croat readers but hackneyed to modern
English readers (Osers 1996; Jones 2000: 78–9).

• nding an analogy (e.g. Scots for the Herzegovinan dialect of Serbo-Croat: Jones 2000: 81). This,
however, may not exactly replicate the source variety's associations.

• shifting to another marked variety, whether along the same axis (e.g. from archaic to hyper-modern:
Holmes 1988: 41) or a di erent axis (e.g. from regional to informal). This almost always changes the
variety's associations.

• shifting to standard language. This avoids the risks of the other approaches, but also removes the
source variety's e ect. When the source poem is ‘multi-voiced’—when changes of variety mark out
 di erent protagonists or di erent ideological viewpoints—it removes this structuring e ect (Jones
2000: 81–2).

Finally, source-culture-speci c associations, references to other works, and the poem's place within its
wider poetic culture may be hard to recreate or analogize (Hron 1997: 18–19; Holmes 1988: 47)—especially
if a poem's restricted format gives no room for explicitation. Hence published translations often supply this
information via an Introduction and/or Translator's Notes.

p. 175 Cognitive processes

Here I summarize ndings from self-reports and think-aloud studies into re-creative poetry translating
(Peraldi 1978, Bly 1983, Honig 1985, Born 1993, McEwan 1991, Doce 1997, Hofstadter 1997, Flynn 2004,

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

Jones 1989, 2006a, 2006b).

Translating poetry is relatively time-consuming, painstaking work. Translations tend to take shape via a
succession of TT ‘versions’: typically, the rst is semantically literal, with later versions bringing in issues
of sound and general poetic e ectiveness. Versions are almost always produced over several drafts, or
working sessions interspersed by ‘rest period[s] in the drawer’ (Born 1993: 61). Each session typically
involves several runs-through of the poem. Units of translating and revising within a run-through typically
correspond to a poem's own subdivision into verses, couplets, lines, and half-lines. Within these units,
translators tackle individual problems (lexis, rhyme, etc.) via strategic ‘micro-sequences’ (Jones 2006b).

Translators read and reread the source poem, target versions, working notes, etc. whilst writing and
rewriting versions and notes: after a rst reading run-through, there is no evidence of separate reading and
writing phases. Translators tend to refer to the source poem at all stages: exclusively TT-oriented runs-
through are rare. Translators are also concerned to reconstruct the poet's intent (about the real-world
inspiration for the poem, say), asking the poet where possible. When choosing translation solutions,
however, they do not necessarily see this as overruling their direct experience of the text as a reader.

Translators spend most time tackling problems of lexis: words and xed expressions. They are also strongly
concerned with underlying poetic image: exploring the source poem's use of imagery, and attempting to
recreate this in the translation. Less translating time is typically spent on sound (rhyme, rhythm,
assonance, etc.), unless translators are trying to recreate formal rhyme and rhythm.

After an initial orienting decision to rhyme, say, or re-create the source rhythm, translating decisions seem
usually made on the spot, according to poetic micro-context rather than overtly voiced principles. In the Li
Po poem, for example, Pound may have weighed up líng-lóng's internal features (its sound features, literal
meaning of ‘jade-tinkling’, combined meaning of ‘exquisite’, etc.) and its structural links (to the image of a
crystal-bead blind, the leitmotif of jade and its connotations of richness, alliteration with the previous word
lián, etc.), and then decided that clear gave the best onomatopoeic, alliterative, and idiomatic link with
crystal curtain. This also means that nal versions rarely fall into one of the archetypes proposed by Lefevere
(phonemic, semantic, metrical, prose, rhyming, etc.: 1975), but are usually hybrid in nature.

Poetry translation is popularly seen as ‘creative’. If we see creative problem-solving as involving solutions
which are both novel and appropriate relative to the source text (Sternberg and Lubart 1999: 3), re-creative
p. 176 translators seem to consider semantically novel solutions only reluctantly and gradually. For example, if
the source poem plays on an idiom's literal and gurative senses (e.g. the Dutch onze handen over ons hart
streken, lit. ‘stroked our hands over our heart’, guratively ‘showed compassion’: Jones 2006a), translators
rst seek solutions that keep all relevant elements (e.g. hands + heart + compassion). If this fails, they
consider solutions that reproduce at least some of the elements (e.g. in English, had a heart). Only if testing
against co-text shows this to be inappropriate do translators consider semantically novel solutions, i.e.
solutions with no ST motivation (e.g. took the plunge)—though even here, loyalty to underlying intent (in
this case, the source poet's explanation that the idiom described relief at moving into a new house) satis es
the appropriateness criterion. If novelty is de ned less strictly as any departure from ST structures,
however, then any ‘adaptive shift’ may be seen as creative: transferring rhymes to other words than those
which bear the rhyme in the ST, for example. Re-creative translators, however, distinguish quite sharply
between semantic novelty (undertaken reluctantly if at all) and adaptive shifts (undertaken as a matter of
course).
 Finally, interviews and post-translation reports show di erences between translators in terms of overall
strategic orientation (e.g. preferring to prioritize sound at the expense of semantic equivalence, or vice
versa). And di erent translators' nal versions of the same source poem can di er radically—especially,
perhaps, if the source poem sets high formal challenges in terms of sound structure, word-play, etc., and
thus o ers no simple or obvious solutions. However, the few think-aloud reports available show that
translators working on the same poem have similar task management styles, problem-solving processes,
and problem hierarchies (most time spent on lexis, closely followed by image, etc.). This also holds for the
same translator tackling di erent poem types (apart from a rise in sound-based micro-sequences for
translations from xed form to xed form), and from di erent language types (e.g. Germanic vs. Slavic).

Motivation and a ective factors

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

Poetry translation is typically done voluntarily: in a translator's free time, without payment or for fees that
rarely compensate for the hours involved. However, publishers exert less deadline pressure than with other
genres (Flynn 2004: 277). With little extrinsic motivation from pay or deadlines, translators need intrinsic
and self-motivation to keep working over a project's lifetime. A ective factors are crucial here. Thus
Flynn's poetry translators (2004: 279) reported that a sense of a nity with their texts was important: they
would refuse commissions to translate works they did not like. Once a project is running, support may come
from interpersonal networks (see below): other members of the project team, fellow poetry translators, and
source-culture enthusiasts for the source poet.

12.3 Transforming poetic text

The main task of poetry translators is to translate. I rst examine this as what might be termed ‘cognitive
habitus’ (a cluster of socially de ned information-processing practices), then in terms of cognitive
challenges and processes, and end by discussing a ective (emotional) factors.

12.3.1 Cognitive habitus

Here we look at how poetry translating processes are conceptualized within poetry production, translation,
p. 172 and consumption ‘ elds’, or loose-knit networks of users, texts and institutions (Inghilleri 2005b: 134–
5)—for example, poetry translators within a certain country, or receptor-language poets. As socially
mediated concepts vary across time and place, I should point out that my claims may be biased towards
recent European practices.

Poetry translation is typically overt. Target readers know they are reading a translator's interpretation of a
source-language poem (Boase-Beier 2004: 25–6). Hence translators may be less free than original poets to
ignore their readers' needs and abilities, and readers may read translated poems more critically than non-
translated poems.

Poetry translators are typically concerned to interpret a source poem's layers of meaning, to relay this
interpretation reliably, and/or to ‘create a poem in the target language which is readable and enjoyable in its
own right, with merit as an independent, literary text’ (Phillips 2001: 23–4; cf. Boase-Beier 2004: 25–6;
Lefevere 1975; Honig 1985: 177; Flynn 2004: 281–2; Jones 2006a). This triple habitus, of course, guides
overt translation in all literary genres. In poetry, however, surface semantics and underlying imagery are
often so closely and complexly bound with linguistic form that it is notoriously di cult to interpret these
relationships and reproduce them in a foreign-language text that meets Phillips's quality demands. This
has inspired a popular discourse of poetry translation as loss. Some lament the loss of source-text
reproduction, as in Lefevere's view that most poetry translations ‘are unsatisfactory renderings of the
source text’ because they fail to capture its totality (1975: 99). Others lament the loss of target-text quality,
as in Robert Frost's reputed saying that ‘poetry is what is lost in translation’ (in Untermeyer 1964: 18). In
practice, translators face a range of output options (Boase-Beier 2009, Hanson 1992). Prose translations
convey only source semantics, usually to help readers read the source (receptor-language poets in
collaborative partnerships, say: see below). What might be called re-creative translations aim to convey not
just the source's complex link between imagery and ‘core sense of the words’ (Reynolds 2003: 108) but also
its poetic e ects, in a viable receptor-culture poem. Rather than trying (and inevitably failing) to nd exact
‘equivalents’ for all a source poem's features, re-creative translators seek ‘counterparts’ and ‘analogues’
 (Holmes 1988: 53–4). Self-reports and source—target text comparisons indicate that most translators
advocate this approach, even those who see themselves primarily as receptor-language poets (see e.g.
Honig 1985; Hughes 1989: 17–18). Some receptor-language poets, however, prefer writing ‘adaptations’
(Mahon 2006, Paterson's ‘versions’, 2006): poems more loosely based on other-language sources.

Re-creative translating is potentially the most challenging approach, for it demands three expertises of the
translator: expert poetry-reading ability in the source language; expert poetry-writing ability in the
receptor language; and mediating between the demands of ST loyalty and TT quality (cf. Keeley 2000: 19).

p. 173 12.3.2 Challenges

Here we look more closely at some of the challenges which re-creative poetry translators face, and the

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

solutions they may choose—whether for deliberate poetic e ect or because they see no better alternative.

First, reading a source poem can involve recognizing and interpreting a highly complex set of meanings and
poetic features. These may even be intentionally obscure—with modernist verse, for example (Bouchard
1993: 149).

According to Boase-Beier, when translating it is crucial to stay true to a source poem's style (its ‘perceived
distinctive manner of expression’: Wales, cited in Boase-Beier 2006: 4), because style encodes the source
writer's attitude towards the content (2004: 28–9). Stylistic loyalty is rarely straightforward, however, as
the following paragraphs show.

With poetic form, Holmes sees translators as choosing between three main approaches (1988: 25–7):

• Mimetic: replicating the original form. This implies openness to the source culture's foreignness
(Holmes 1988: 25–6). However, the form may carry di erent weight in the receptor culture (Hejinian
1998, Ra el 1988)—a ve-syllable line feels ‘classical’ in Chinese, for example, but may seem radically
compressed in French.

• Analogical: using a target form with a similar cultural function to the source form (e.g. the English
iambic pentameter for the Chinese ve-syllable line). This implies a belief that receptor-culture
poetics has universal value (Holmes 1988: 26).

• Organic: choosing a form that best suits the translator's ‘own authenticity’ of response to the source
(Scott 1997: 35). This stresses the impossibility of recreating the source form—content link (Holmes
1988: 28).

According to Holmes (1988: 54), nding close correspondence between source and target is easier with ‘a
poem that leans very close to prose’, but the more complex the poem, the bigger the compromises that
translators have to make. This may mean that sub-genres of poetry di er in di culty: narrative verse, say,
may give the translator more room for manoeuvre than the more compressed lyric verse (Davis 1996: 31–2).

Some translators, particularly into languages with a strong free-verse tradition, advocate free-verse
translations of xed-form (rhyme and/or rhythm-based) source poems, often on analogical or organic
grounds. Others argue that this risks losing crucial stylistic e ects. Thus, regarding Hughes's free-verse
renderings of Pilinszky's Hungarian verse, Csokits writes: ‘without the softening e ect of the original metre
and rhyme scheme […] they sound harsher and Pilinszky's view of the world appears grimmer’ (1989: 11).

p. 174 Similarly, attitudes towards ST rhyme range from abandonment to re-creation (users 1998), with partial
preservation (replacing full by half-rhyme, say) as a compromise. Key arguments for abandonment are:

• Rhyme may have negative associations (e.g. old-fashioned or trite) for receptor-genre readers.

• Finding rhyme-words is di cult, especially when the receptor language has less exible word order
and/or a greater variety of word-endings than the source (e.g. English relative to German and Italian
respectively: Osers 1996; Feldman 1997: 5).

• Seeking rhyme leads to unacceptable semantic shifts—such as having ‘to add images that destroy the
poem's integrity’ (Bly 1983: 44–5).
 Those who advocate recreating a source poem's rhyme scheme, whether mimetically or analogically, admit
that this requires technical skill, but argue that rhyme is an integral part of the poem's meaning: ‘if one
disapproves of rhyme in poetry, one should not translate poems that rhyme’ (Barnstone 1984:50–51; cf.
Mo ett 1989, 1999). Moreover, though seeking rhymes may give radical shifts in surface wording, the
underlying images can be preserved (Jones 2007).

Source poets may deliberately use ‘marked’ language varieties: language that, relative to the standard
variety, is distinctively archaic or modern, informal or formal, regional, speci c to poetry or typical of other
genres, or simply idiosyncratic. Alternatively, language varieties that might have seemed unmarked to the
poet may appear non-standard to most modern readers. Translators then face a choice between:

• replicating the source variety. This may not always replicate its e ect, however: archaisms, for
example, may seem original and exciting to modern Serbo-Croat readers but hackneyed to modern

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

English readers (Osers 1996; Jones 2000: 78–9).

• nding an analogy (e.g. Scots for the Herzegovinan dialect of Serbo-Croat: Jones 2000: 81). This,
however, may not exactly replicate the source variety's associations.

• shifting to another marked variety, whether along the same axis (e.g. from archaic to hyper-modern:
Holmes 1988: 41) or a di erent axis (e.g. from regional to informal). This almost always changes the
variety's associations.

• shifting to standard language. This avoids the risks of the other approaches, but also removes the
source variety's e ect. When the source poem is ‘multi-voiced’—when changes of variety mark out
di erent protagonists or di erent ideological viewpoints—it removes this structuring e ect (Jones
2000: 81–2).

Finally, source-culture-speci c associations, references to other works, and the poem's place within its
wider poetic culture may be hard to recreate or analogize (Hron 1997: 18–19; Holmes 1988: 47)—especially
if a poem's restricted format gives no room for explicitation. Hence published translations often supply this
information via an Introduction and/or Translator's Notes.

p. 175 12.3.3 Cognitive processes

Here I summarize ndings from self-reports and think-aloud studies into re-creative poetry translating
(Peraldi 1978, Bly 1983, Honig 1985, Born 1993, McEwan 1991, Doce 1997, Hofstadter 1997, Flynn 2004,
Jones 1989, 2006a, 2006b).

Translating poetry is relatively time-consuming, painstaking work. Translations tend to take shape via a
succession of TT ‘versions’: typically, the rst is semantically literal, with later versions bringing in issues
of sound and general poetic e ectiveness. Versions are almost always produced over several drafts, or
working sessions interspersed by ‘rest period[s] in the drawer’ (Born 1993: 61). Each session typically
involves several runs-through of the poem. Units of translating and revising within a run-through typically
correspond to a poem's own subdivision into verses, couplets, lines, and half-lines. Within these units,
translators tackle individual problems (lexis, rhyme, etc.) via strategic ‘micro-sequences’ (Jones 2006b).

Translators read and reread the source poem, target versions, working notes, etc. whilst writing and
rewriting versions and notes: after a rst reading run-through, there is no evidence of separate reading and
writing phases. Translators tend to refer to the source poem at all stages: exclusively TT-oriented runs-
through are rare. Translators are also concerned to reconstruct the poet's intent (about the real-world
inspiration for the poem, say), asking the poet where possible. When choosing translation solutions,
however, they do not necessarily see this as overruling their direct experience of the text as a reader.

Translators spend most time tackling problems of lexis: words and xed expressions. They are also strongly
concerned with underlying poetic image: exploring the source poem's use of imagery, and attempting to
recreate this in the translation. Less translating time is typically spent on sound (rhyme, rhythm,
assonance, etc.), unless translators are trying to recreate formal rhyme and rhythm.

After an initial orienting decision to rhyme, say, or re-create the source rhythm, translating decisions seem
usually made on the spot, according to poetic micro-context rather than overtly voiced principles. In the Li
Po poem, for example, Pound may have weighed up líng-lóng's internal features (its sound features, literal
 meaning of ‘jade-tinkling’, combined meaning of ‘exquisite’, etc.) and its structural links (to the image of a
crystal-bead blind, the leitmotif of jade and its connotations of richness, alliteration with the previous word
lián, etc.), and then decided that clear gave the best onomatopoeic, alliterative, and idiomatic link with
crystal curtain. This also means that nal versions rarely fall into one of the archetypes proposed by Lefevere
(phonemic, semantic, metrical, prose, rhyming, etc.: 1975), but are usually hybrid in nature.

Poetry translation is popularly seen as ‘creative’. If we see creative problem-solving as involving solutions
which are both novel and appropriate relative to the source text (Sternberg and Lubart 1999: 3), re-creative
p. 176 translators seem to consider semantically novel solutions only reluctantly and gradually. For example, if
the source poem plays on an idiom's literal and gurative senses (e.g. the Dutch onze handen over ons hart
streken, lit. ‘stroked our hands over our heart’, guratively ‘showed compassion’: Jones 2006a), translators
rst seek solutions that keep all relevant elements (e.g. hands + heart + compassion). If this fails, they
consider solutions that reproduce at least some of the elements (e.g. in English, had a heart). Only if testing

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

against co-text shows this to be inappropriate do translators consider semantically novel solutions, i.e.
solutions with no ST motivation (e.g. took the plunge)—though even here, loyalty to underlying intent (in
this case, the source poet's explanation that the idiom described relief at moving into a new house) satis es
the appropriateness criterion. If novelty is de ned less strictly as any departure from ST structures,
however, then any ‘adaptive shift’ may be seen as creative: transferring rhymes to other words than those
which bear the rhyme in the ST, for example. Re-creative translators, however, distinguish quite sharply
between semantic novelty (undertaken reluctantly if at all) and adaptive shifts (undertaken as a matter of
course).

Finally, interviews and post-translation reports show di erences between translators in terms of overall
strategic orientation (e.g. preferring to prioritize sound at the expense of semantic equivalence, or vice
versa). And di erent translators' nal versions of the same source poem can di er radically—especially,
perhaps, if the source poem sets high formal challenges in terms of sound structure, word-play, etc., and
thus o ers no simple or obvious solutions. However, the few think-aloud reports available show that
translators working on the same poem have similar task management styles, problem-solving processes,
and problem hierarchies (most time spent on lexis, closely followed by image, etc.). This also holds for the
same translator tackling di erent poem types (apart from a rise in sound-based micro-sequences for
translations from xed form to xed form), and from di erent language types (e.g. Germanic vs. Slavic).

12.3.4 Motivation and a ective factors

Poetry translation is typically done voluntarily: in a translator's free time, without payment or for fees that
rarely compensate for the hours involved. However, publishers exert less deadline pressure than with other
genres (Flynn 2004: 277). With little extrinsic motivation from pay or deadlines, translators need intrinsic
and self-motivation to keep working over a project's lifetime. A ective factors are crucial here. Thus
Flynn's poetry translators (2004: 279) reported that a sense of a nity with their texts was important: they
would refuse commissions to translate works they did not like. Once a project is running, support may come
from interpersonal networks (see below): other members of the project team, fellow poetry translators, and
source-culture enthusiasts for the source poet.

p. 177
Non-translating tasks

Poetry translators may do other textual tasks besides translating poems. They may act as editors: choosing
the poems for a translated selection, say; or even, as with pre-modern texts, establishing a de nitive source
text (Crisafulli 1999: 83 .).

Translators may write paratextual material: a critical preface and/or endnotes (or, more rarely, footnotes).
These typically supply background information about the source poet, work, and context, though they may
also describe translating approach, decisions, and points of source/target di erence (Crisafulli 1999; Bishop
2000: 66–7). Translators may also give public readings from their translations, often with the source poet
reading the source poems.
p. 177
12.4 Non-translating tasks

Poetry translators may do other textual tasks besides translating poems. They may act as editors: choosing
the poems for a translated selection, say; or even, as with pre-modern texts, establishing a de nitive source
text (Crisafulli 1999: 83 .).

Translators may write paratextual material: a critical preface and/or endnotes (or, more rarely, footnotes).
These typically supply background information about the source poet, work, and context, though they may
also describe translating approach, decisions, and points of source/target di erence (Crisafulli 1999; Bishop
2000: 66–7). Translators may also give public readings from their translations, often with the source poet
reading the source poems.

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

Teams

The translator and other players

A poetry translation project involves not just a translator, but a multi-person production team, which may
include a print or web publisher, editor, source poet, graphic artist, etc. (Jones 2009). Translators, however,
are not always the most powerful actors in terms of making production decisions and recruiting others into
the project. Thus a project may be initiated by a publisher or editor, who commissions one or more
translators to translate, or who requests existing translations from translators or source poets. This is a
typical pattern for multi-poet anthologies (Jones 2009). With works of a single living source poet, the poet
may initiate and control the project, especially if s/he lives in the country of publication.

Sometimes, however, translators may initiate a project by seeking a publisher, whether independently
(when translating a dead poet) or on the request of a source poet; the latter often happens when the
translator lives in the publisher's country and the source poet does not (Jones 2009). The Dutch-native
poetry translators interviewed by Flynn (2004: 279) reported that successfully nding a publisher is
‘directly proportionate’ to the translator's reputation, but also that ‘a company's publishing policy is
in uenced by translators’. And once the project is under way, translators may well liaise with editors or
publishers on behalf of the source poet or his/her agents—especially, again, if the translations are
appearing in the translator's home country.

When it comes to actual translating decisions, poetry translators are often allowed considerable autonomy
p. 178 by other team players. Flynn's translators report that publishers or editors rarely give translators an
explicit translating brief, especially if translators have a high reputation: they assume that translators
‘know what should be done’, and usually request no more than minor textual decisions (Flynn 2004: 280).
After submission, copy-editors make fewer corrections to poetry than to literary-prose translations (p.
278). Some source poets who read the receptor language, however, may insist on approving all textual
decisions; and if they disagree with the translator, their higher social capital may mean that their opinion
prevails, even if TT quality su ers (Keeley, in Honig 1985: 148–9; Weissbort 2004).

Poetry translators may well translate from more than one language and national literature—like all of
Flynn's interviewees (2004: 276). Restrictions on SL knowledge or reading and writing skills can be
overcome by collaborative translating. A typical pattern is where an expert reader of the source language
works with a native writer of the receptor language (e.g. Kunitz and Weissbort 1989, Csokits 1989, Hughes
1989). The former may also be the source poet. If the SL reader is primarily a linguist and the TL writer a
published poet, however, people outside the team may see the former as of lower status: ‘translators of
literals […] are the pariahs of the realm of letters’ (Csokits 1989: 14). Another common collaborative pattern
is where both translators are SL readers and TL writers, but feel that shared expertise or complementary
working styles lead to better results (e.g. Keeley 2000: 32–7). Poets may also translate each other (Lesser
1989)—even in the same volume, as in Paz and Tomlinson's Spanish—English poetic correspondence
Airborn/Hijos del Aire (1981).

‘Multi-agency’, in fact, is a reality of professional poetry translating (Flynn 2004: 277)—perhaps, again,
because of the complexity of poetic communication. Even solo translators may consult ST informants about
the ST (about unknown words and ‘references’, say: p. 277), and ask target-version readers to advise on
 output quality (Bly 1983: 42–3). Potential ST informants may be the poet, if living (Kline 1989), or other
native readers of the source literature. Potential target-version readers may or may not know the source
language (Bishop 2000: 65); some translators argue that not being able to read the source allows readers to
focus more clearly on TL draft quality. Fellow translators may play both roles (Flynn 2004: 276–7).

Positionality
Positionality indicates where a project player's allegiance lies (Toury 1980, cited in Tymoczko 2003: 184). It
may be seen in terms of physical location, but also of a ective loyalty (Jones 2009). Poetry translation
teams very often have a distributed positionality. In other words, a team's players are typically located in
both SL and TL countries, and even third countries; and Internet publication means that readers may be
anywhere in the world, especially if the TL is an international lingua franca. The players' loyalty tends to be

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

p. 179 primarily to the source poet, culture, or point of view they aim to communicate. But as their implicit brief
is always to communicate a poetic message to receptor readers, there is also a loyalty towards those readers:
this may mean not only making the message comprehensible but also, very often, communicating the
project's cultural or ideological aim to readers (see e.g. the introduction to Agee 1998).

12.5 Teams

12.5.1 The translator and other players

A poetry translation project involves not just a translator, but a multi-person production team, which may
include a print or web publisher, editor, source poet, graphic artist, etc. (Jones 2009). Translators, however,
are not always the most powerful actors in terms of making production decisions and recruiting others into
the project. Thus a project may be initiated by a publisher or editor, who commissions one or more
translators to translate, or who requests existing translations from translators or source poets. This is a
typical pattern for multi-poet anthologies (Jones 2009). With works of a single living source poet, the poet
may initiate and control the project, especially if s/he lives in the country of publication.

Sometimes, however, translators may initiate a project by seeking a publisher, whether independently
(when translating a dead poet) or on the request of a source poet; the latter often happens when the
translator lives in the publisher's country and the source poet does not (Jones 2009). The Dutch-native
poetry translators interviewed by Flynn (2004: 279) reported that successfully nding a publisher is
‘directly proportionate’ to the translator's reputation, but also that ‘a company's publishing policy is
in uenced by translators’. And once the project is under way, translators may well liaise with editors or
publishers on behalf of the source poet or his/her agents—especially, again, if the translations are
appearing in the translator's home country.

When it comes to actual translating decisions, poetry translators are often allowed considerable autonomy
p. 178 by other team players. Flynn's translators report that publishers or editors rarely give translators an
explicit translating brief, especially if translators have a high reputation: they assume that translators
‘know what should be done’, and usually request no more than minor textual decisions (Flynn 2004: 280).
After submission, copy-editors make fewer corrections to poetry than to literary-prose translations (p.
278). Some source poets who read the receptor language, however, may insist on approving all textual
decisions; and if they disagree with the translator, their higher social capital may mean that their opinion
prevails, even if TT quality su ers (Keeley, in Honig 1985: 148–9; Weissbort 2004).

Poetry translators may well translate from more than one language and national literature—like all of
Flynn's interviewees (2004: 276). Restrictions on SL knowledge or reading and writing skills can be
overcome by collaborative translating. A typical pattern is where an expert reader of the source language
works with a native writer of the receptor language (e.g. Kunitz and Weissbort 1989, Csokits 1989, Hughes
1989). The former may also be the source poet. If the SL reader is primarily a linguist and the TL writer a
published poet, however, people outside the team may see the former as of lower status: ‘translators of
literals […] are the pariahs of the realm of letters’ (Csokits 1989: 14). Another common collaborative pattern
is where both translators are SL readers and TL writers, but feel that shared expertise or complementary
working styles lead to better results (e.g. Keeley 2000: 32–7). Poets may also translate each other (Lesser
 1989)—even in the same volume, as in Paz and Tomlinson's Spanish—English poetic correspondence
Airborn/Hijos del Aire (1981).

‘Multi-agency’, in fact, is a reality of professional poetry translating (Flynn 2004: 277)—perhaps, again,
because of the complexity of poetic communication. Even solo translators may consult ST informants about
the ST (about unknown words and ‘references’, say: p. 277), and ask target-version readers to advise on
output quality (Bly 1983: 42–3). Potential ST informants may be the poet, if living (Kline 1989), or other
native readers of the source literature. Potential target-version readers may or may not know the source
language (Bishop 2000: 65); some translators argue that not being able to read the source allows readers to
focus more clearly on TL draft quality. Fellow translators may play both roles (Flynn 2004: 276–7).

12.5.2 Positionality

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

Positionality indicates where a project player's allegiance lies (Toury 1980, cited in Tymoczko 2003: 184). It
may be seen in terms of physical location, but also of a ective loyalty (Jones 2009). Poetry translation
teams very often have a distributed positionality. In other words, a team's players are typically located in
both SL and TL countries, and even third countries; and Internet publication means that readers may be
anywhere in the world, especially if the TL is an international lingua franca. The players' loyalty tends to be
p. 179 primarily to the source poet, culture, or point of view they aim to communicate. But as their implicit brief
is always to communicate a poetic message to receptor readers, there is also a loyalty towards those readers:
this may mean not only making the message comprehensible but also, very often, communicating the
project's cultural or ideological aim to readers (see e.g. the introduction to Agee 1998).

Second-order networks

In analysing social interaction, Milroy (1987: 46–7) distinguishes between close-knit ‘ rst-order
networks’ (like the translation project teams just described) and larger, looser ‘second-order networks’.
Two of the latter are discussed here: the poetry translation profession and the ‘communities’ (user groups
etc.) with an ‘interest’ in a project (Venuti 2000b: 477). As a profession involves a shared sense of identity,
institutions, etc., it may also be seen as a Bourdieusian ‘ eld’. And as communities of interest involve not
only people but also the texts they read and write, they may also involve the textual networks known as
‘systems’ (Hermans 1999).

Profession
Though published poetry translators rarely work full-time for full pay, they may in other respects be
regarded as professionals: they have a special expertise which is valued and recognized by those who use
their services, and which they are usually allowed to use autonomously (cf. Freidson 1994: 210).

Professions tend to be distinguished by their own institutions and habitus (ways of behaving sanctioned by
the network: Inghilleri 2005b: 134–5). Institutionally, poetry translators are only weakly professionalized.
Professional accreditation of poetry translators, therefore, is almost always informal, by word-of-mouth
recommendation and reputation among second-order networks of literary production (publishers, editors,
poets, fellow translators, etc.).

As for poetry translators' professional habitus, we have already discussed cognitive aspects. It is, however,
worth mentioning status and visibility. Poetic messages are typically seen as: canonical, i.e. highly valued by
the community that uses them; requiring high expertise to create, because of their complex textual features;
and requiring autonomous, individualized working practices, because of their special communicative
purposes. These, plus an awareness that poetic messages rarely have one-to-one equivalents across
languages, mean that poetry translation is also seen by many users as requiring high, autonomously
p. 180 wielded expertise—even impossibly high, as shown in the popular discourse of translation loss. This
helps to explain why published poets often translate poetry—as opposed to novels, say, which are rarely
translated by published novelists. As a result, poetry translators often enjoy higher status than translators
of other genres. With some (such as Robert Bly in the US or Octavio Paz in Mexico), this may derive partially
or largely from their reputation as receptor-language poet. If the source poet is internationally known, even
‘non-poet’ translators may acquire some of the source poet's status by dint of association (Keeley 2000:
 104). And if their translations are judged successful, even non-poet translators of previously unknown poets
may acquire respect among communities of readers.

This status is recognized and stimulated by two practices common in poetry but unusual in other genres,
which give the poetry translator high visibility. One is the prominent display of the translator's name: on a
book cover, say, or beneath poems in multi-translator anthologies. Another, mentioned above, is that
poetry translators often write paratextual materials to accompany their translations.

Professional ethics form a key aspect of habitus. Translators' accounts show a strong ethic of loyal
representation among poetry translators: a desire to communicate the ‘essence’ of the source poem to
target readers in the most e ective means possible (Bly 1983: 30–31; Barnstone 1984: 50). This, of course,
underlies not only poetry translation but also professional translation in other genres. Even producers of
poetry-to-prose translations and adaptations may be seen as subscribing to this ethic: believing that poetry

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

translation cannot be both semantically reliable and poetically e ective, they aim to convey loyally
whatever aspect of the source poem they see as most relevant to their communicative purpose (its
semantics, or its poetic e ect). As be ts a non-institutionalized profession, there are few if any
quali cations or degree programmes in poetry translation. Training is largely informal and self-driven,
with poets or linguists gradually developing an interest and (often, though not always) expertise in poetry
translation through practice. Some other-directed training does happen, however. Universities, translators'
and writers' associations may run poetry translation workshops, and poetry translation modules may be
o ered within creative writing or translator-training degree programmes and extra-mural courses.

Communities of interest and systems

Venuti's ‘community of interest’ refers to the network of general readers, critics, TL poets, etc. a ected by a
translation project (2000b: 477). Conditions of translation and reception may a ect which texts are o ered
to a reader community. With English translations of Eastern European poetry during the Cold War, for
example, translators tended to select ‘translatable’ poets, constructing a stereotype of ‘Eastern European
p. 181 poetry’ as metaphor-oriented political poetry in free verse, which in turn conditioned the choice of
further translations (Jarniewicz 2002b; Sampson 2001: 83). If a project does not t the receptor culture's
expectations about domestic or translated poetry, by contrast, it may not be accepted by receptor readers
and critics, or its reception may not accord with the production team's aims (Malroux 1997: 20; Flynn 2004:
279). There are similar risks if receptor readers have no knowledge about the source poet(s) or their literary
culture (Dutch poetry in the 1980s UK, for instance: Holmes 1988: 12–13)—a knowledge which
Introductions typically aim to supply.

Receptor-language poets often see their own output as in uenced by translated poetry, whether as
translators or readers—Octavio Paz in Mexico or Ted Hughes in the UK, for instance (Dumitrescu 1995;
Jarniewicz 2002a: 93). Output may even extend to pseudo-translations (original poems which claim to be
translations) and poetry that makes deliberate use of translationese, as with Christopher Reid's
‘translations’ of the imaginary poet Katerina Brac (1985; cf. Jarniewicz 2002a: 93–5). Such in uences may
extend to domestic poetry systems as a whole: ‘the translation of foreign poetry can be a means of
revitalizing our own poetry’ (Mao [1922]2004). And they can stimulate ‘trans-linguistic’ literary
movements (Paz 1973, cited in Dumitrescu 1995: 240).

There are other potential communities of interest: those in the source country who wish to see ‘their’ poet
published, for instance. Here, by deciding whether or not to translate, poetry translators may play a
gatekeeping role, controlling the poet's access to a wider community or even (with a globalized TL) a global
community of readers. When translation does happen, however, it often con rms or enhances the poet's
status at home.

Communities of interest might also be trans-national—those within and outside Bosnia supporting the
anti-nationalist motives of the Scar on the Stone anthology, for instance (Agee 1998). Communities of
interest may interact with other communities: Scar on the Stone's reader community, for example, might
interact with the wider community of UK poetry readers, and in opposition to communities within and
outside Bosnia which support ethno-nationalist models of politics and culture.
 12.6 Second-order networks

In analysing social interaction, Milroy (1987: 46–7) distinguishes between close-knit ‘ rst-order
networks’ (like the translation project teams just described) and larger, looser ‘second-order networks’.
Two of the latter are discussed here: the poetry translation profession and the ‘communities’ (user groups
etc.) with an ‘interest’ in a project (Venuti 2000b: 477). As a profession involves a shared sense of identity,
institutions, etc., it may also be seen as a Bourdieusian ‘ eld’. And as communities of interest involve not
only people but also the texts they read and write, they may also involve the textual networks known as
‘systems’ (Hermans 1999).

12.6.1 Profession

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

Though published poetry translators rarely work full-time for full pay, they may in other respects be
regarded as professionals: they have a special expertise which is valued and recognized by those who use
their services, and which they are usually allowed to use autonomously (cf. Freidson 1994: 210).

Professions tend to be distinguished by their own institutions and habitus (ways of behaving sanctioned by
the network: Inghilleri 2005b: 134–5). Institutionally, poetry translators are only weakly professionalized.
Professional accreditation of poetry translators, therefore, is almost always informal, by word-of-mouth
recommendation and reputation among second-order networks of literary production (publishers, editors,
poets, fellow translators, etc.).

As for poetry translators' professional habitus, we have already discussed cognitive aspects. It is, however,
worth mentioning status and visibility. Poetic messages are typically seen as: canonical, i.e. highly valued by
the community that uses them; requiring high expertise to create, because of their complex textual features;
and requiring autonomous, individualized working practices, because of their special communicative
purposes. These, plus an awareness that poetic messages rarely have one-to-one equivalents across
languages, mean that poetry translation is also seen by many users as requiring high, autonomously
p. 180 wielded expertise—even impossibly high, as shown in the popular discourse of translation loss. This
helps to explain why published poets often translate poetry—as opposed to novels, say, which are rarely
translated by published novelists. As a result, poetry translators often enjoy higher status than translators
of other genres. With some (such as Robert Bly in the US or Octavio Paz in Mexico), this may derive partially
or largely from their reputation as receptor-language poet. If the source poet is internationally known, even
‘non-poet’ translators may acquire some of the source poet's status by dint of association (Keeley 2000:
104). And if their translations are judged successful, even non-poet translators of previously unknown poets
may acquire respect among communities of readers.

This status is recognized and stimulated by two practices common in poetry but unusual in other genres,
which give the poetry translator high visibility. One is the prominent display of the translator's name: on a
book cover, say, or beneath poems in multi-translator anthologies. Another, mentioned above, is that
poetry translators often write paratextual materials to accompany their translations.

Professional ethics form a key aspect of habitus. Translators' accounts show a strong ethic of loyal
representation among poetry translators: a desire to communicate the ‘essence’ of the source poem to
target readers in the most e ective means possible (Bly 1983: 30–31; Barnstone 1984: 50). This, of course,
underlies not only poetry translation but also professional translation in other genres. Even producers of
poetry-to-prose translations and adaptations may be seen as subscribing to this ethic: believing that poetry
translation cannot be both semantically reliable and poetically e ective, they aim to convey loyally
whatever aspect of the source poem they see as most relevant to their communicative purpose (its
semantics, or its poetic e ect). As be ts a non-institutionalized profession, there are few if any
quali cations or degree programmes in poetry translation. Training is largely informal and self-driven,
with poets or linguists gradually developing an interest and (often, though not always) expertise in poetry
translation through practice. Some other-directed training does happen, however. Universities, translators'
and writers' associations may run poetry translation workshops, and poetry translation modules may be
o ered within creative writing or translator-training degree programmes and extra-mural courses.
 12.6.2 Communities of interest and systems
Venuti's ‘community of interest’ refers to the network of general readers, critics, TL poets, etc. a ected by a
translation project (2000b: 477). Conditions of translation and reception may a ect which texts are o ered
to a reader community. With English translations of Eastern European poetry during the Cold War, for
example, translators tended to select ‘translatable’ poets, constructing a stereotype of ‘Eastern European
p. 181 poetry’ as metaphor-oriented political poetry in free verse, which in turn conditioned the choice of
further translations (Jarniewicz 2002b; Sampson 2001: 83). If a project does not t the receptor culture's
expectations about domestic or translated poetry, by contrast, it may not be accepted by receptor readers
and critics, or its reception may not accord with the production team's aims (Malroux 1997: 20; Flynn 2004:
279). There are similar risks if receptor readers have no knowledge about the source poet(s) or their literary
culture (Dutch poetry in the 1980s UK, for instance: Holmes 1988: 12–13)—a knowledge which

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

Introductions typically aim to supply.

Receptor-language poets often see their own output as in uenced by translated poetry, whether as
translators or readers—Octavio Paz in Mexico or Ted Hughes in the UK, for instance (Dumitrescu 1995;
Jarniewicz 2002a: 93). Output may even extend to pseudo-translations (original poems which claim to be
translations) and poetry that makes deliberate use of translationese, as with Christopher Reid's
‘translations’ of the imaginary poet Katerina Brac (1985; cf. Jarniewicz 2002a: 93–5). Such in uences may
extend to domestic poetry systems as a whole: ‘the translation of foreign poetry can be a means of
revitalizing our own poetry’ (Mao [1922]2004). And they can stimulate ‘trans-linguistic’ literary
movements (Paz 1973, cited in Dumitrescu 1995: 240).

There are other potential communities of interest: those in the source country who wish to see ‘their’ poet
published, for instance. Here, by deciding whether or not to translate, poetry translators may play a
gatekeeping role, controlling the poet's access to a wider community or even (with a globalized TL) a global
community of readers. When translation does happen, however, it often con rms or enhances the poet's
status at home.

Communities of interest might also be trans-national—those within and outside Bosnia supporting the
anti-nationalist motives of the Scar on the Stone anthology, for instance (Agee 1998). Communities of
interest may interact with other communities: Scar on the Stone's reader community, for example, might
interact with the wider community of UK poetry readers, and in opposition to communities within and
outside Bosnia which support ethno-nationalist models of politics and culture.

Researching poetry translation

Finally, it is worth looking at poetry's role in translation studies research.

Poetry accounts for a tiny proportion of world translation output. Case studies and examples taken from
poetry, however, have dominated theory-building in translation studies at the expense of more frequently
p. 182 translated genres—even recently, as with Venuti's domestication/foreignization discussions, which are
based largely on his own poetry translation practice (1995). One reason might be the rich variety of
problems o ered by poetry translation. Another might be that literary translation, including poetry
translation, can give rich information about cultural and inter-cultural ideologies and ‘interfaces’ (Lefevere
1975: 111–20; Tymoczko 1999: 30). There are risks in over-extending theories inspired by poetry translation
into genres with very di erent communicative rules, such as technical translation.

There is room, however, to research poetry translation in its own right, perhaps as part of a wider aim to
map novice and expert translation across genres. Our knowledge about poetry translating is, perhaps
surprisingly, still fragmentary. Many studies have compared speci c source and target texts, and many
after-the-event reports about how poetry translators tackled speci c works. However, these rarely
generalize beyond the individual case, and are hard to compare. There have been no book-length surveys of
poetry translation as a whole, at least in English, since the 1970s (Lefevere 1975, De Beaugrande 1978). And
the more rigorous research methods that have recently done much to map non-literary translation have
hardly been applied to poetry. Few published studies using structured translator interviews or think-alouds
look at poetry (apart from, say, Flynn 2004 and Jones 2006b). And I know of no concordance studies into
 poetry translation, or ethnographic accounts of poetry translation projects (contrast e.g. Buzelin 2006 and
Koskinen 2008).

12.7 Researching poetry translation

Finally, it is worth looking at poetry's role in translation studies research.

Poetry accounts for a tiny proportion of world translation output. Case studies and examples taken from
poetry, however, have dominated theory-building in translation studies at the expense of more frequently
p. 182 translated genres—even recently, as with Venuti's domestication/foreignization discussions, which are
based largely on his own poetry translation practice (1995). One reason might be the rich variety of

Downloaded from https://academic.oup.com/edited-volume/34526/chapter/292916492 by University of Manchester user on 28 February 2023

problems o ered by poetry translation. Another might be that literary translation, including poetry
translation, can give rich information about cultural and inter-cultural ideologies and ‘interfaces’ (Lefevere
1975: 111–20; Tymoczko 1999: 30). There are risks in over-extending theories inspired by poetry translation
into genres with very di erent communicative rules, such as technical translation.

There is room, however, to research poetry translation in its own right, perhaps as part of a wider aim to
map novice and expert translation across genres. Our knowledge about poetry translating is, perhaps
surprisingly, still fragmentary. Many studies have compared speci c source and target texts, and many
after-the-event reports about how poetry translators tackled speci c works. However, these rarely
generalize beyond the individual case, and are hard to compare. There have been no book-length surveys of
poetry translation as a whole, at least in English, since the 1970s (Lefevere 1975, De Beaugrande 1978). And
the more rigorous research methods that have recently done much to map non-literary translation have
hardly been applied to poetry. Few published studies using structured translator interviews or think-alouds
look at poetry (apart from, say, Flynn 2004 and Jones 2006b). And I know of no concordance studies into
poetry translation, or ethnographic accounts of poetry translation projects (contrast e.g. Buzelin 2006 and
Koskinen 2008).

Further reading and relevant resources

The posthumously published Translated! Papers on Literary Translation and Translation Studies (1988) by
James Holmes—a poetry translator and a founding father of academic translation studies—remains crucial
reading. Weissbort (1989) and Allén (1999) present insightful collections of articles written by both poetry
translators and scholars. Honig (1985) provides fascinating reports of interviews with poetry translators,
and Bly (1983) describes in detail his own processes of translating and revising a poem. Key overviews of
more speci c aspects are given by Osers (rhyme and rhythm: 1996, 1998) and Boase-Beier (style and reader
cognition: 2004). Flynn (2004) gives a useful survey of poetry translation in its wider professional context,
and Jones (2006b) presents a framework for research into poetry translating processes.

Further reading and relevant resources

The posthumously published Translated! Papers on Literary Translation and Translation Studies (1988) by
James Holmes—a poetry translator and a founding father of academic translation studies—remains crucial
reading. Weissbort (1989) and Allén (1999) present insightful collections of articles written by both poetry
translators and scholars. Honig (1985) provides fascinating reports of interviews with poetry translators,
and Bly (1983) describes in detail his own processes of translating and revising a poem. Key overviews of
more speci c aspects are given by Osers (rhyme and rhythm: 1996, 1998) and Boase-Beier (style and reader
cognition: 2004). Flynn (2004) gives a useful survey of poetry translation in its wider professional context,
and Jones (2006b) presents a framework for research into poetry translating processes.