After you load a translation memory either from TMX files or from Total Recall, CafeTran performs an automatic matching. The currently-translated segment is matched automatically against this translation memory to produce the results of matching. As your translation memory is getting bigger and bigger over time and reaches thousands of segments, the automatic matching may get slower as the current source segment is matched against that large amount of translation memory segments. To overcome this issue of huge memories, you can perform the preliminary memory matching of the whole source document so that the matches will be available instantly, without any delay. The preliminary memory matching may take a while but you do not have to wait until it completes because matches are available right after the segment is processed. It means that you can translate at your own human pace while the preliminary matching is taking place in the background.


  1. Load segments to the working memory either from a TMX file or a Total Recall table.
  2. Choose one of the Preliminary memory matching commands in the Translation menu:
  • Preliminary matching from current segment.
  • Preliminary matching all segments.
  • Preliminary matching untranslated segments.

Alternatively, you can set the preliminary matching for a translation memory before it is opened. In the memory options panel, select "Preliminary memory matching" from the "Workflow integration" drop-down list.