Poetry Translating as Expert Action
Wholist processors lessen cognitive load by tackling the overall picture without
overloading working memory with detail. Thus Carl reported starting with two
word-processor copies of a source sonnet side-by-side on screen, one of which he
transferred line-by-line into the receptor language, aiming for a quick version of
the whole text: “I work like a sketch artist, striving for a good likeness”. He then
gradually refined this version so that it resembled the source as closely as possible
in microstructural and macrostructural terms: hence he did not ignore detail, but
tackled it in a whole-text framework (“what I am seeing is always fourteen lines, so
that I haven’t got this feeling that I am building something up”).
Other interviewees cited both analytic and wholist processing, typically at dif-
ferent stages. Thus, with Alan’s co-translating project, work was initially wholist:
his source-language-linguist co-translator “introduced” him to the source poet
and then read out the source poem several times. Both co-translators then
discussed analytically the source poem’s sounds, “shape”, and “patterns”, plus the
“formality or informality of the language, how colloquial it is, how formal, how
difficult, how complex”. Alan later wrote this information into a working version,
which was jointly polished in further analytic question-and-answer sessions. Fi-
nally, they asked for wholist feedback: “we would show the English version to one
or two people to see how it reads to them as an English poem”.4.3.3.3 Early-version wording
Some interviewees said that their early versions recorded ‘alternative solutions’ to
be chosen from later, whereas others said that they had the ‘pure line’ of a proto-
poem. Though this resembles the analytic ↔ wholist continuum, it is not the same:
one translator (Ellen) is strongly analytic in organizing her translation work, but
prefers pure-line early versions. Hence a separate ‘Early-Version Wording’ con-
tinuum was posited: see Figure 18.
Thus the alternative-solution Bruce reported often producing several early
versions, one carrying “simple meaning” (semantic content), and another the
“shape of the poem” (presumably poetic form), with either or both giving alterna-
tive word-meanings, rhymes, etc.; only later are these used to generate what looks
like a target poem. The pure-line Carl and Ellen, by contrast, worked with one al-
ternative-free target version from the outset, gradually refining this by word-proc-
essor overtyping or using pencil and rubber.Alternative solutions Pure line
Bruce Derek Alan Carl
Ellen