Chapter 5. Five translators translate
Kouwenaar, however, added: “I’m only saying that as a guide and you don’t have to
account for that in the translation”.^2
Because my pilot drafts and the other translators’ main-study drafts of Toen wij
were gathered under very similar conditions, they were ultimately merged into one
five-translator dataset – both to enrich the data and, as explained above, to give a
foundation for Chapter 6’s analyses. The only real difference is that I, having finished
translating before the workshop, could not use Kouwenaar’s input, whereas the oth-
er translators could. This, however, is not enough to invalidate my data: if nothing
else, both I and the other translators used other source-poem informants besides the
source poet. For transparency’s sake, my own data (‘Francis’) are not anonymized.
All five end-of-Draft 3 versions are shown in Figure 24.
Recordings were transcribed and think-alouds coded by other researchers.
Transcriptions and codings were then checked and edited by myself.5.2.2 Validity
In Chapters 5 and 6, I am again both translating subject and researcher. This gives
a rich insider view – for instance, my having translated Toen wij gives in-depth
understanding of the challenges which the other translators faced. To reduce the
risk of the subject role infecting the researcher role with Toen wij, however, I took
two safeguards besides those mentioned earlier (p. 36 ff.). Firstly, though timing
pressures meant I had to record Fleur’s Draft 1 between my Drafts 1 and 2, I gath-
ered the remaining other-translator data after my own Draft 3, thus minimizing the
chance of others’ translating decisions influencing mine. In the few places where I
felt they did so, I stated this in the think-alouds, where it was logged as Text-helper
input (see Figure 27: p. 226). Secondly, I analysed the four translators’ data before
my own, so that their data formed the initial norm to which mine was later com-
pared. The two analyses were only merged, for readability’s sake, at a late stage.
The methodological pros and cons of think-alouds are to some extent the con-
verse of translator interviews. Think-alouds give detailed data about real translat-
ing. However, so much data is generated and transcribing recordings is so time-
consuming that one study can observe relatively few translators or texts, making
generalisation harder (Bernardini 2001). Translators’ real-time running commen-
taries tally very closely with their translating processes; being conscious and fo-
cused on the here-and-now, however, they cannot access what is not explicitly
conscious, such as automatized processes or implicit reasons for actions (ibid.).
Off-line data outside translation proper are also hard to capture: Bernardini cites
“social interaction” here, though the think-alouds in Chapters 5 and 6 did capture- “en dat zeg ik alleen als handleiding en dat hoef je niet te verantwoorden in de vertaling.”