The Workflow of Computer-Assisted Translation Tools in

4
thick arrows in fi gure 1 indicate where the source text stands at each of the three stages. These arrows
indicate how some of the tasks performed affect the source text directly, such as ‘Term extraction’,
whereas the benefi ts of other tasks, such as ‘Text alignment’, are more indirect.
In addition to exemplifying the workfl ow in chronological order, this division into three stages
is also intended to highlight the particular importance of the ‘Before translation’ stage. The quality
and quantity of the tasks performed during this stage have a direct bearing on how effective
the translation with CAT tools is in the ‘During translation’ stage. This is further explored in the
following section and the ‘During translation’ and ‘After translation’ stages are discussed in sections
3.2 and 3.3 respectively.
3.2. Preparing for translation
When undertaking any translation, preparation work is always essential, but in the context of
working with CAT tools, the tasks performed prior to translation take on special signifi cance because
the higher the quality and quantity of the tasks performed during this stage, the more effi -
cient the translation of the source text will be afterwards. This is because the two main databases
common to most CAT tools, the translation memory and the terminological database (or termbase),
as explained in section 1.2 above, are initially empty when starting to work with CAT tools.
It is up to the CAT tool user to populate both the translation memory and the termbase with relevant
material prior to translation, whenever this is possible.
The main tasks that can be performed prior to translation are text alignment and terminology
extraction, but some CAT tools also offer the possibility of converting previous or third-party
databases into standardised formats in the translation industry such as .tbx, which allows the exchange
of data between CAT tools. The results of text alignment (typically sentences) will be fed
to the translation memory and the results of terminology extraction (typically terms) will be fed
into the termbase (or terminological database).