formulaic expressions per 1,000 words, as we have already seen. This is a relatively
high average that should not be ignored, as an especially large number of errors were
made by the translators when translating many formulaic expressions. It is bad enough
to make a mistake in ORDINARY translation, but in technical translation, for example,
‘[c]onfusing directions accompanying an essential device in a jet plane may result in
disaster’ (Freedman 1958: 10).
From the classification of formulaic expressions, another conclusion to be drawn
is that there is little correlation between the errors within subclasses. For instance, the
percentage of errors in the variable classes was 28.8%, whilst the same percentage in the
invariable classes was 29.6%. However, there is a marked difference between the main
subclasses: the percentage of errors for sentential/pseudo-sentential classes was very
low (4.3%) compared to the total percentage of errors in the sub-sentential classes
(29.8%). This means that the translation of both classes of formulaic expressions is
problematic for translators, but translators must direct their efforts to improving the
quality of the translation towards the translation of sub-sentential classes much more
than to the sentential classes.
The occurrence of a formulaic expression in the corpus should not prompt the
translator to find an exact equivalent in the target language, especially as there might
not be one, but rather to re-read the whole paragraph and select the best possible
translation, even if it means that the formulaic expression in question is not translated
with another formulaic expression in the target language. Formulaic expressions in the
source text do not necessarily require formulaic expressions in the target text.
This leads us to issue of identifying formulaic expressions. Formulaic
expressions may not need formulaic translations but translators do need to recognize
them as formulaic in order to render them into the target text with the appropriate
meaning and identifying formulaic expressions in a text is a difficult task with no magic
solutions at hand. Most of the errors made by the translators of this corpus seem to
indicate that the formulaic expressions in the source text were not recognized as such in
the source texts by the translators, with a resulting infelicitous translation at best and an
ungrammatical and therefore meaningless translation at worst.