Poetry Translating as Expert Action
English speech, then he worked it afterwards into poetry”^22. Moreover, “learned
the book” almost certainly describes the pre-reading and analysis performed be-
fore Phase 1 by this study’s translators. This suggests that the pre-reading → liter-
al-version → poetic-version sequence is universally common, even if not always
followed (Carl, for instance, claims that he skips the literal-version stage).
During a run-through, the poem is processed as one or more textual chunks,
each of the maximum size that can be held in working memory while it is worked
on during a ‘macro-sequence’ of operations. Chunks range in size from clauses or
Lines (especially with slow runs-through), via stanzas, to the whole poem
(especially with quick runs-through). With chunks shorter than the whole poem,
translators work through the poem sequentially, thus ensuring that no potentially
problematic text is missed; but they may also backtrack, to establish or check how
individual chunks fit with broader structures of meaning or style.5.4.1.2 Establishing and conveying meaning
Within macro-sequences, strategic and non-strategic micro-sequences tackle dis-
crete translation problems and manage operations respectively. Most micro-se-
quence work focuses on exploring and (re)constructing textual meaning, within
four clusters of processes.
Firstly, Lexis work, sometimes backed up by Grammar/Discourse and sound
work, establishes microstructures of source-poem meaning (semantic, intrinsic-
poetic, stylistic and intertextual). These are transferred into corresponding target-
language microstructures by recording literal equivalents, synonyms and notes.
Secondly, Image work shapes macrostructures of source-poem meaning and
builds them into a coherent source-text world. This combines two processes
(cf. Kussmaul, in Dancette and Ménard 1996: 142). The low-level microstructures
from Process 1 are used ‘bottom-up’ to build macrostructures, i.e. higher-level
meaning schemata. But these macrostructures, together with higher-level sche-
mata about the poem’s extra-textual context (the source poet’s account of the po-
em’s inspiration, for example), give the translator extra ‘top-down’ help in shaping
or fine-tuning new macrostructures from their underlying microstructures.
Thirdly, translators can then check whether source and target text worlds
differ significantly at microstructural and macrostructural level – and if so, try
to reduce the gap by rewriting target-version microstructures. Where the source
poem highlights linguistic features specific to the source language, finding
equivalent target-language microstructures becomes problematic. With To e n
wij, this happened mainly with multi-meaning items – such as reactivated- “þās bōc hæfde geleornode ond of Lædene tō Engliscum spelle gewende, þā geworhtē hē hī
eft tō lēoðe” (Anonymous author c890/1967: 9).