Forum: CAT Tools Technical Help
Topic:"Apply the TM breakdown"
Poster: Christine Andersen
Post title: ALWAYS measure against the client's TM or a blank TM
If you give a client an estimate like that, always use their TM or an empty TM. I can then run some texts against my own TM and find some more matches, but that does not concern the client. (The matches I find in my TM may or may not be suitable in their context, or they may simply be standard phrases like 'Page 3 of 6' and the like.)
It probably depends a bit on your language pair how much you benefit from fuzzy matches, too.
I work with Danish as the source and English as my target language - there are not a lot of inflections etc. to check in the source, so the chances are that matches are quite useful. Even then, I never manage to save as much time as some agencies claim. The 'overhead time' of setting up, checking and keeping TMs up to date is not significantly affected, even if the time between starting to translate and delivering the job is shortened to some extent.
For me, anything over 75% is normally quite useful, and I even like the AutoSuggest and Fragment Match features in Trados... But if you work with languages where you also have to check for gender or other agreements in all your fuzzy matches, then even the very high percentages may or may not be worth your while, I'm not sure.
We seriously have to fight the 'one size fits all' approach to translation. Each text is different, and although you can define some general rules and patterns, you cannot be sure they apply to your languages and the specific text you are looking at.
Topic:"Apply the TM breakdown"
Poster: Christine Andersen
Post title: ALWAYS measure against the client's TM or a blank TM
If you give a client an estimate like that, always use their TM or an empty TM. I can then run some texts against my own TM and find some more matches, but that does not concern the client. (The matches I find in my TM may or may not be suitable in their context, or they may simply be standard phrases like 'Page 3 of 6' and the like.)
It probably depends a bit on your language pair how much you benefit from fuzzy matches, too.
I work with Danish as the source and English as my target language - there are not a lot of inflections etc. to check in the source, so the chances are that matches are quite useful. Even then, I never manage to save as much time as some agencies claim. The 'overhead time' of setting up, checking and keeping TMs up to date is not significantly affected, even if the time between starting to translate and delivering the job is shortened to some extent.
For me, anything over 75% is normally quite useful, and I even like the AutoSuggest and Fragment Match features in Trados... But if you work with languages where you also have to check for gender or other agreements in all your fuzzy matches, then even the very high percentages may or may not be worth your while, I'm not sure.
We seriously have to fight the 'one size fits all' approach to translation. Each text is different, and although you can define some general rules and patterns, you cannot be sure they apply to your languages and the specific text you are looking at.
